Numerical control device, learning device, inference device and operation screen display method for a numerical control device

Abstract

A numerical control device (1) comprises a learning data acquisition unit that receives learning data including operation log information and operation state information, and a model generation unit that, using the learning data, generates a learned model for inferring screen display data for displaying a new operation screen from the operation log information and the operation state information. The operation log information represents a user's operation performed on an operation screen of the numerical control device (1) that controls a machine tool and represents information of the operation screen used in the operation. The operation state information represents a condition of the machine tool at a time when the operation shown in the operation log information is performed.The new operation screen includes information searched by the user and extracted from several existing operation screens.

Description

Technical area

[0001] The present disclosure relates to a numerical control device, a learning device, an inference device, and an operation screen display method for a numerical control device. background

[0002] A numerical control device that controls a machine tool includes an operation screen that is used in a wide variety of situations. The operation screen must be user-friendly depending on the usage situation. For example, Patent Literature 1 discloses a numerical control device capable of rearranging menu items on an operation screen depending on the operator using machine learning. Citation listPatent literature

[0003] Patent Literature 1: Japanese Patent No. 6898479 Overview of the inventionTechnical problem

[0004] However, with the previous conventional technology, when the information the user is seeking is located on multiple operation screens, the user must find the required information from the multiple screens in one user operation. This still poses a problem of complexity in user operation.

[0005] The present disclosure has been made in view of the foregoing, and it is an object of the present disclosure to provide a numerical control device that enables improvement of the user's work efficiency even when the information searched by the user exists in a plurality of operation screens in existing operation screens. Solution to the problem

[0006] In order to solve the problems described above and achieve the object, a numerical control device according to the present disclosure includes: a learning data acquisition unit for obtaining learning data including operation log information and operation state information, wherein the operation log information represents a user's operation performed on an operation screen of the numerical control device that controls a machine tool and also represents information of the operation screen used in the operation, wherein the operation state information represents a condition of the machine tool at a time when the operation shown in the operation log information is performed;and a model generation unit for generating a learned model using the learning data for inference of screen display data for displaying a new operation screen from the operation log information and the operation state information, wherein the new operation screen includes information extracted from a plurality of pre-existing operation screens; Advantageous effects of the invention

[0007] A numerical control device according to the present disclosure offers an advantage that it can provide a numerical control device that enables improvement of the user's work efficiency even when the information searched by the user exists in a plurality of operation screens in existing operation screens. Short description of the drawings Fig. 1 is a diagram showing a functional configuration of a numerical control device according to a first embodiment. Fig. 2 is a diagram showing a detailed configuration of a learning device associated with the numerical control device for a machine tool. Fig. 3 is a flowchart concerning a learning process of the learning device. Fig. 4 is a configuration diagram of an inference device associated with the numerical control device for a machine tool. Fig. 5 is a flowchart concerning an inference process of the inference device. Fig. 6 is an explanatory diagram relating to an advantage of the first embodiment. Fig. 7 is a diagram showing an example of a new operation screen displayed by the numerical control device according to a third embodiment. Fig. 8 is a diagram showing a functional configuration of a numerical control device according to a fourth embodiment. Fig. 9 is a diagram showing an example of a method for distributing a learned model in a fifth embodiment. Fig. 10 is a diagram for describing functionality of the numerical control device according to a sixth embodiment. Fig. 11 is a diagram showing an example of a new operation screen displayed by the numerical control device according to a seventh embodiment. Fig. 12 is a diagram showing a functional configuration of a numerical control device according to an eleventh embodiment. Fig. 13 is a diagram showing a dedicated hardware element for implementing the functionality of the numerical control devices according to the first to eleventh embodiments. Fig. 14 is a diagram showing a configuration of a control circuit for implementing the functionality of the numerical control devices according to the first to eleventh embodiments. Description of the embodiments

[0008] A numerical control device, a learning device, an inference device, and an operation screen display method for a numerical control device according to embodiments of the present disclosure will be described in detail below with reference to the drawings. First embodiment.

[0009] Fig. 1 is a diagram showing a functional configuration of a numerical control device 1 according to a first embodiment. The numerical control device 1 includes a control unit 11, an operation log acquisition unit 12, a data acquisition unit 13, a learning device 14, a learned model storage unit 15, an inference device 16, and an output unit 17. The numerical control device 1 is a device that controls a machine tool according to a machining program. The numerical control device 1 controls a movement of the machine tool by controlling a drive unit 2 included in the machine tool. The drive unit 2 includes an amplifier, a motor, and the like. The numerical control device 1 is also capable of controlling a peripheral device 3.Furthermore, the numerical control device 1 can output an operation screen for operating the numerical control device 1 using a display device 4. It should be noted that in . Fig. 1, the display device 4 is shown as a separate device from the numerical control device 1, but the display device 4 may be included in the numerical control device 1. Fig. 1 also shows an input device 5 as a separate device from the numerical control device 1, but the input device 5 may be included in the numerical control device 1.

[0010] The control unit 11 is connected to the drive unit 2 and the peripheral device 3, which are controlled by the numerical control device 1. The control unit 11 has the functionality to control the amplifier and the motor included in the drive unit 2 and has the functionality to control the peripheral device 3. The control unit 11 receives operation state information from each of the drive unit 2 and the peripheral device 3, representing an operation state of a corresponding one of these devices, and outputs the obtained operation state information to the data acquisition unit 13.The operation status information may include, for example, an operation mode such as "machining," "setting," or "alarm issued," a status of a signal from a programmable logic controller (PLC), a status of a machining program for controlling the drive unit 2 of the machine tool, and the like. When the operation mode is "alarm issued," the operation status information may also include the type of alarm, details of the alarm, and the like.

[0011] The operation log acquisition unit 12 obtains operation log information representing an operation performed by the user on an operation screen of the numerical control device 1. The operation log information includes information of the operation screen used in the operation. The operation screen information may include, for example, identification information for identifying the operation screen, hierarchy information representing a relationship to another operation screen when operation screens are hierarchically arranged, screen display data for displaying the operation screen, and the like. The operation log acquisition unit 12 may obtain the operation log information based on input information received from the input device 5.The input device 5 is, for example, a keyboard for receiving keyboard input, a pointing device for receiving a specific coordinate position on an operation screen, a microphone for receiving voice input, and / or the like. Note that the pointing device is, for example, a mouse, a touch sensor, and / or the like. The touch sensor may be a touch panel that integrally includes the input device 5 and the display device 4, or a separate touchpad.

[0012] The data acquisition unit 13 receives various types of data used by the numerical control device 1. The data acquisition unit 13 is capable of outputting the obtained data to the learning device 14 and the inference device 16. The data acquisition unit 13 can receive operation log information from the operation log acquisition unit 12 and operation status information from the control unit 11.

[0013] The learning device 14 generates a learned model for inference from screen display data for displaying a new operation screen of the numerical control device 1 using machine learning. The learning device 14 stores the learned model, which is a result of the learning, in the learned model storage unit 15. The inference device 16 infers screen display data for displaying a new operation screen using the learned model and outputs the screen display data, which is a result of the inference, to the output unit 17.

[0014] The output unit 17 outputs a new operation screen using the display device 4 based on the screen display data output from the inference device 16. In addition, the screen display data may include at least one type of information, namely the design of the new operation screen, the information output timing, an information output method, and an operation guide, which is information for assisting the user in operation. The design of the operation screen may include information about the position on the screen for displaying the information searched by the user, the size of an area for displaying the information searched by the user, and the like. The information output timing provides the user with information about the timing of information output.The information output method represents highlighting, animated representation, and whether a screen transition occurs. The operation guide is information that assists the user in operation, such as additional information for information output to the user and how to operate the screen. Note that generating new screen display data enables the display of a new operation screen. Accordingly, generating new screen display data may be referred to hereinafter as "generating a new operation screen." Note that the numerical control device 1 can generate a new operation screen using the inference device 16 at a time such as when the operation mode of the machine tool changes or when the user performs an operation on the operation screen. <lernphase>

[0015] Fig. Figure 2 is a diagram showing a detailed configuration of the learning device 14 associated with the numerical control device 1 for a machine tool. The learning device 14 includes a learning data acquisition unit 141 and a model generation unit 142. The learning data acquisition unit 141 receives the

[0016] Operation log information and operation status information as learning data.

[0017] The model generation unit 142 learns screen display data based on the learning data including the operation log information and the operation state information. That is, the model generation unit 142 generates a learned model for inferring screen display data for displaying a new operation screen from the operation log information and the operation state information of the numerical control device 1 for a machine tool. Note that the "new operation screen" includes information extracted from a plurality of existing operation screens.In this regard, the term "information extracted from multiple existing operation screens" refers to information obtained based on the operation log information and the operation state information as information searched by the user in the operation log information, such as information displayed by the user in a previous operation.Note that the term "information extracted from multiple existing operation screens" means searching for extracted "information" from information included in multiple existing operation screens, and applies not only to the case where the thus extracted information is information separately included in multiple existing operation screens, but also to the case where the thus extracted information is included in a single operation screen among the existing operation screens. Furthermore, the "new operation screen" includes information that contributes to improving the work efficiency of the work to be performed with the operation screen. The "information that contributes to improving work efficiency" includes information for reducing the user's operation burden.Examples of user operation burden include the number of operations performed, the number of screen transitions, and operation time. Examples of information to reduce user operation burden include information extracted from an existing operation screen, information that includes a combination of information extracted from existing operation screens, and operation manuals. Furthermore, "information contributing to improving work efficiency" includes, for example, information that has a positive effect on an operation result. Examples of information that has a positive effect on an operation result include information for preventing user operation errors, such as operation manuals.

[0018] The model generation unit 142 may use a learning algorithm, which is a well-known algorithm, such as supervised learning, unsupervised learning, and reinforcement learning. In the following description, a case of using reinforcement learning is described as an example. In reinforcement learning, an agent, which is an actor in a certain environment, observes a parameter of the environment that represents the current state and determines an action to be taken. An action taken by the agent causes a dynamic change in the environment, and the agent receives a reward that depends on the change in the environment. The agent repeats this to learn an action strategy that maximizes the reward to be achieved through a series of actions. Q-learning and TD learning techniques are known as typical reinforcement learning techniques.For example, in the case of Q-learning, a typical update formula of an action value function Q(s, a) is expressed by the mathematical formula (1). Formula 1: Q(st,at)←Q(st,at)+α(rt+1+γmax︸aQ(st+1,a))−Q(st,at)

[0019] In the mathematical formula (1) s represents t the state of the environment at time t and a t the action at time t. The action a t causes the transition of the state to a state s t+1 . The parameter r t+2 represents the reward given by this state transition, γ represents a discount factor, and α represents a learning rate. Note that γ takes a value of 0<γ≤1 and α takes a value of 0<α≤1. An optimal action a t in state s t at time t is compared with the operation log information associated with action a t and the operating status information corresponding to the state s t correspond, learned.

[0020] The update formula expressed by mathematical formula (1) increases the action value Q when the action value Q of an action "a" having a highest Q value at time t+1 is greater than the action value Q of an action "a" at time t, and decreases the action value Q in the opposite case. In other words, the action value function Q(s, a) is updated to make the action value Q of an action "a" at time t approach an optimal action value at time t+1. This causes the optimal action value in a certain environment to propagate sequentially into action values ​​in previous environments.

[0021] When a learned model is to be generated by reinforcement learning as described above, the model generation unit 142 comprises a reward calculation unit 143 and a function update unit 144 as shown in Fig. 2 shown.

[0022] The reward calculation unit 143 calculates a reward based on the information of the operation log and the information of the operation state. The reward calculation unit 143 calculates a reward "r" based on reward criteria. The reward criteria include a reward increase criterion for increasing the reward "r" and a reward decrease criterion for decreasing the reward "r." In this example, the reward criteria may each be a criterion based on information representing the user's work efficiency, such as at least one of the number of operation executions, the number of screen transitions, or the operation time, because the reward "r" is preferably increased or decreased depending on the user's work efficiency.The number of operation executions, the number of screen transitions, and the operation time are counted on a per-operation mode basis during the operation of the machine tool. For example, a reward of "1" is given to increase the reward "r" as the number of operation executions decreases, and a reward of "-1" is given to decrease the reward "r" as the number of operation executions increases. Alternatively, the reward "r" may be increased as the number of screen transitions decreases, and the reward "r" may be decreased as the number of screen transitions increases. As a further alternative, the reward "r" may be increased as the operation time decreases, and the reward "r" may be decreased as the operation time increases.Note that in the above description, it is assumed that the decision of whether to increase or decrease the reward is made based on one of the number of operation executions, the number of screen transitions, and the operation time. However, the decision may be made based on two of the number of operation executions, the number of screen transitions, and the operation time, or based on all of the number of operation executions, the number of screen transitions, and the operation time. Furthermore, the decision of whether to increase or decrease the reward "r" may be made based on information other than the number of operation executions, the number of screen transitions, and the operation time.It is sufficient for the reward calculation unit 143 to increase the reward "r" when it can be determined that the user's work efficiency is increasing, and to decrease the reward "r" when it can be determined that the user's work efficiency is decreasing.

[0023] The function updating unit 144 updates a function for determining the screen display data according to the reward "r" calculated by the reward calculating unit 143 and outputs the resulting function to the learned model storage unit 15. For example, in a case of Q-learning, the action value function Q(s t , a t ) is used as the function to calculate the screen display data.

[0024] The learning device 14 executes the above learning operation iteratively. The learned model storage unit 15 stores the action value function Q(s t , a t ) which is updated by the feature updating unit 144, ie the learned model.

[0025] A learning process performed by the learning device 14 will next be described with reference to Fig. 3 described. Fig. 3 is a flowchart relating to a learning process of the learning device 14.

[0026] The learning data acquisition unit 141 obtains operation log information and operation state information as learning data (step S10).

[0027] The model generation unit 142 calculates a reward "r" based on the operation log information and the operation state information (step S11). Specifically, the reward calculation unit 143 may calculate the reward "r" using, for example, predetermined reward criteria. The reward criteria include a reward increase criterion for increasing the reward "r" and a reward decrease criterion for decreasing the reward "r." As described above, each of the reward criteria may be a criterion based on information representing the user's work efficiency, such as at least one of the number of operation executions, the number of screen transitions, or the operation time.For example, when reward criteria based on the number of operation executions are used, the reward calculation unit 143 counts the number of operation executions on a per operation mode basis based on the operation log information and the operation state information, and determines that the reward increase criterion is satisfied when the number of operation executions decreases, and determines that the reward decrease criterion is satisfied when the number of operation executions increases.

[0028] If the reward calculation unit 143 determines that the reward increase criterion is met (step S11: reward increase criterion), the reward calculation unit 143 increases the reward "r" (step S12). If the reward calculation unit 143 determines that the reward decrease criterion is met (step S11: reward reduction criterion), the reward calculation unit 143 reduces the reward "r" (step S13).

[0029] The function update unit 144 updates the action value function Q(s t , a t ) based on the reward "r" calculated by the reward calculation unit 143 (step S14), wherein the action value function Q(s t , a t ) is expressed by the mathematical formula (1) and is stored in the learned model storage unit 15.

[0030] The learning device 14 iteratively executes the preceding steps from step S10 to step S14 and stores the generated action value function Q(s t , a t ) than the learned model.

[0031] Although the learning device 14 according to the present embodiment has been described as storing the learned model in the learned model storage unit 15 provided outside the learning device 14, the learned model storage unit 15 may be included in the learning device 14. <nutzungsphase>

[0032] Fig. Figure 4 is a configuration diagram of the inference device 16 associated with the numerical control device 1 for a machine tool. The inference device 16 includes an inference data acquisition unit 161 and an inference unit 162.

[0033] The inference data acquisition unit 161 receives the operation log information and the operation state information.

[0034] The inference unit 162 infers screen display data using the learned model. That is, inputting the operation log information and the operation state information obtained by the inference data acquisition unit 161 into this learned model enables the inference of screen display data suitable for the operation log information and the operation state information.

[0035] It should be noted that, although the present embodiment has been described in which the screen display data is output using a learned model obtained by learning performed by the learning device 14 included in the numerical control device 1 for a machine tool, this operation may be performed in such a manner that a learned model is obtained from another numerical control device 1 and the screen display data is output based on this learned model.

[0036] A method for obtaining the screen display data using a learned model will be described next with reference to Fig. 5 described. Fig. 5 is a flowchart relating to an inference process of the inference device 16.

[0037] The inference data acquisition unit 161 obtains the operation log information and the operation state information as data for inference (hereinafter referred to as inference data) (step S20).

[0038] The inference unit 162 inputs the operation log information and the operation state information into the learned model stored in the learned model storage unit 15 (step S21) and obtains the screen display data. The inference unit 162 outputs the obtained screen display data to the output unit 17 (step S22).

[0039] The numerical control device 1 causes the display device 4 to display an operation screen using the output screen display data (step S23). The operation screen displayed in this phase is a new operation screen that includes user-requested information extracted from multiple existing operation screens. This allows the user to combine multiple pieces of user-requested information into a single new operation screen, even if these multiple pieces of information are included separately on multiple screens in the existing operation screens.

[0040] Note that, although the inference unit 162 described using reinforcement learning as the learning algorithm in the present embodiment, the learning technique is not limited to this. In addition to reinforcement learning, a learning technique such as supervised learning, unsupervised learning, or semi-supervised learning may also be used as the learning algorithm.

[0041] Additionally, the learning algorithm to be used in the model generation unit 142 may be deep learning, in which the extraction of a functionality is learned by itself. Machine learning may otherwise be performed using other known methods, such as neural networks, genetic programming, functional logic programming, support vector machines, or the like.

[0042] It should be noted that the learning device 14 and the inference device 16 may be connected to the numerical control device 1, for example, via a network, and may be separate devices from this numerical control device 1. Otherwise, the learning device 14 and the inference device 16 may be integrated into the numerical control device 1, as shown in Fig. 1. Furthermore, the learning device 14 and the inference device 16 can be provided in a cloud server.

[0043] In addition, the model generation unit 142 can learn screen display data using learning data obtained from a plurality of numerical control devices 1. Note that the model generation unit 142 can obtain learning data from a plurality of numerical control devices 1 used in the same area, or learn the screen display data using learning data collected from a plurality of numerical control devices 1 operating independently in different areas. Furthermore, a numerical control device 1 for collecting learning data can be added to or removed from the numerical control devices 1 during operation.In addition, the learning device 14 that has learned screen display data associated with a specific numerical control device 1 can be used by another numerical control device 1 different from the specific numerical control device 1 to relearn and update screen display data associated with the other numerical control device 1.

[0044] Next, we describe operation screens generated using a learned model. Fig. 6 is an explanatory diagram relating to an advantage of the first embodiment. As shown in the upper part of Fig. As shown in Figure 6, the existing operation screens are arranged hierarchically. Existing screen #1 and existing screen #2 are located immediately below the upper screen. Furthermore, existing screen #1-1, existing screen #1-2, and existing screen #1-3 are located immediately below existing screen #1. Existing screen #2-1 and existing screen #2-2 are located directly below existing screen #2. This example assumes that the information sought by the user is information #1, information #2, and information #3. It is also assumed that information #1 is included in existing screen #1-1, information #2 is included in existing screen #1-3, and information #3 is included in existing screen #2-2.In this case, when using only the existing operation screens, the user needs to perform operation (1) to transition from the upper screen to the existing screen #1, and then perform operation (2) to transition from the existing screen #1 to the existing screen #1-1 to reach information #1. The user then needs to perform operation (3) to return from the existing screen #1-1 to the existing screen #1, and then perform operation (4) to transition from the existing screen #1 to the existing screen #1-3 to reach information #2.The user then needs to further perform the operation (5) of returning from the existing screen #1-3 to the existing screen #1, the operation (6) of transitioning from the existing screen #1 to the top screen, the operation (7) of transitioning from the top screen to the existing screen #2, and then the operation (8) of transitioning from the existing screen #2 to the existing screen #2-2 to reach information #3. That is, to reach all of the information 1, 2, and 3 from the top screen, a total of eight screen transitions are required. In contrast, the numerical control device 1 according to the first embodiment is capable of displaying the generated screen #1, which is a new operation screen including information #1, #2, and #3. This allows the user to display the various information they are looking for by performing one screen transition operation.

[0045] A concrete example of an operation screen to be generated by the numerical control device 1 is described below. For example, if an error has occurred in the machine tool or the peripheral device 3, the user is likely to first check the state of the device in which the error occurred and / or details of the error, and then determine a measure to correct the error. Thus, a first example assumes a case where an alarm indicating a machine tool error was generated during machining. In the first example, an operation screen is generated as a new operation screen in which the user can see, for example, the state of the machine tool, detailed information about the alarm, and the measure to be taken in conjunction with the details of the alarm.In this case, the "condition of the machine tool" corresponds to "Information #1", the "detailed information of the alarm" corresponds to "Information #2", and the "action to be taken in connection with the details of the alarm" corresponds to "Information #3" in the example of . Fig. 6. Alternatively, in a second example, a case is assumed where communication with the peripheral device 3 fails. In the second example, an operation screen is generated as a new operation screen, in which the operation screen allows the user to view information such as the status of the machine tool and the peripheral device 3, the communication status such as the number of communication errors in communication with the peripheral device 3, and the action to be taken. In this case, the "statuses of the machine tool and the peripheral device 3" correspond to "Information #1," the "communication condition such as the number of communication errors in communication with the peripheral device 3" corresponds to "Information #2," and the "action to be taken" corresponds to "Information #3" in the example of Fig. 6.

[0046] Note that in a case where the information required to determine the information sought by the user is insufficient when an error occurs, troubleshooting can be performed interactively. In a third example, when the user inputs a problem using the input device 5, such as a keyboard or microphone, using natural language, the numerical control device 1 is capable of generating a new operation screen depending on the input text or voice and displaying the new operation screen on the display device 4.For example, if the user inputs "I want to back up the entire parameter set," the numerical control device 1 generates an operation screen including, for example, information the user needs to solve the problem, such as a parameter backup screen, an operation manual, and the like, and causes the display device 4 to display the operation screen. The user's interactive operation can be performed iteratively to clarify the problem step by step.

[0047] The first, second, and third examples described above were described with respect to an operation screen when a fault has occurred in a machine tool or the like. A fourth example assumes normal operation without the occurrence of a fault, in which the numerical control device 1 generates a new operation screen including optimal information extracted from existing operation screens of the numerical control device 1 according to what the user wants to do, and causes the display device 4 to display the new operation screen. For example, when the user inputs what they want to do via the input device 5 such as a keyboard or microphone, the numerical control device 1 generates a new operation screen that allows the user to achieve the goal based on the functions of the numerical control device 1.

[0048] As described above, the numerical control device 1 according to the first embodiment includes the learning data acquisition unit 141 that obtains learning data including operation log information and operation state information; and the model generation unit 142 that, using the learning data, generates a learned model for inferring screen display data for displaying a new operation screen from the operation log information and the operation state information, where the operation log information represents a user's operation performed on an operation screen of the numerical control device 1 that controls a machine tool and also represents information of the operation screen used in the operation;The operation state information represents the state of the machine tool at a time when the operation represented in the operation log information was executed; and the new operation screen includes information extracted from a plurality of existing operation screens. Additionally, the numerical control device 1 includes the inference data acquisition unit 161 that obtains inference data including operation log information and operation state information; the inference unit 162 that, using a learned model, outputs screen display data for displaying a new operation screen from the operation log information and the operation state information obtained by the inference data acquisition unit 161;and the output unit 17 that causes the display device 4 to output the new operation screen based on the screen display data, wherein the operation log information represents a user operation performed on an operation screen of the numerical control device 1 that controls a machine tool, and also represents information used in an operation; the operation state information represents the state of the machine tool at a time when the operation represented in the operation log information was performed;and the learned model is a learned model for inferring, from the operation log information and the operation state information, the screen display data for displaying the new operation screen, including information extracted from a plurality of existing operation screens. This makes it possible to provide the user with a user-friendly operation screen according to the operation state of the machine tool. For example, when the information searched by the user is displayed in a plurality of screens in the existing operation screens, extracting the information searched by the user to generate a new operation screen can reduce the number of operation performances, the number of screen transitions, the operation time, and / or the like for the user, as described above with reference to FIG. Fig. 6. This can improve the user's work efficiency. In addition, a method of preparing operation screens in advance, each depending on a factor such as the operation state of the machine tool, can increase the number of previously prepared operation screens, considering that a large number of combinations may occur with respect to the information sought by the user depending on the state of the machine tool. Accordingly, generating a new operation screen based on the operation log information and the operation state information, as performed by the numerical control device 1, can reduce the number of previously prepared operation screens, which can result in an advantage in reducing memory consumption.

[0049] Note that the new operation screen displayed using the screen display data inferred using the learned model can collectively display multiple pieces of information separately included in multiple existing operation screens. This is made possible by the operation log information obtained as learning data, which represents not only the operations performed on a single operation screen but also a history of the operations performed by the user on multiple existing operation screens. For example, when the multiple existing operation screens are arranged hierarchically, the operation log information represents a history of the user's operations performed on the multiple existing operation screens arranged hierarchically.

[0050] In addition, the model generation unit 142 includes the reward calculation unit 143 and the function update unit 144. The reward calculation unit 143 increases the reward "r" when at least one of the number of operation executions, the number of screen transitions, or the operation time decreases, and decreases the reward "r" when at least one of the number of operation executions, the number of screen transitions, or the operation time increases. This allows the learned model to infer from the screen display data the display of a new operation screen that decreases at least one of the following: the number of operation executions, the number of screen transitions, or the operation time.

[0051] It should be noted that the new screen display data generated by the numerical control device 1 includes at least one kind of information related to the design of the new operation screen, the timing of information output, an information output method, and the operation guide, that is, information that helps the user to operate. Second embodiment.

[0052] A second embodiment differs from the first embodiment in the information used as input by the learning device 14 and the inference device 16. The numerical control device 1 has a basic configuration similar to that of the numerical control device 1 of the first embodiment shown in Fig. 1. Therefore, the Fig. 1, and differences from the first embodiment will be mainly described.

[0053] The operation log acquisition unit 12 obtains, as operation log information, time-series data such as a coordinate position input by the user using a pointing device such as a touch panel, a key code input from a keyboard, identification information of an operation screen used in the operation, and / or a hierarchical structure of operation screens. By using such operation log information, the numerical control device 1 can generate a learned model capable of inferring screen display data for displaying a new operation screen designed to reduce the number of times at least one of scrolling, zooming, or reducing operations is performed.For example, the reward calculation unit 143 of the model generation unit 142 increases the reward "r" as the number of dynamic operations such as scrolling, zooming in, and / or out decreases, and decreases the reward "r" as the number of dynamic operations increases. This allows the information sought by the user to be displayed in a size and arrangement that allows for a reduction in the number of dynamic operations.

[0054] As described above, the operation log information according to the second embodiment includes a time series history of coordinate positions input by the user. Additionally, the learned model infers screen display data for displaying a new operation screen designed to reduce the number of operations of at least one of scrolling, zooming in, or zooming out. This can reduce not only the number of screen transitions but also the number of operations performed by the user within a single operation screen, further improving the user's work efficiency. Third embodiment.

[0055] A third embodiment will be described with respect to a method for receiving feedback on a new operation screen. The numerical control device 1 has a basic configuration similar to the numerical control device 1 of the first embodiment. Therefore, the Fig. 1, and the following description mainly describes the differences from the first embodiment.

[0056] In the third embodiment, the new operation screen generated by the numerical control device 1 using the inference device 16 includes a feedback receiving field for receiving user feedback on this operation screen.

[0057] Fig. Fig. 7 is a diagram showing an example of a new operation screen displayed by the numerical control device 1 according to the third embodiment. Fig. The operation screen shown in Figure 7 comprises a first area for displaying detailed alarm information, a second area for displaying machine tool status information representing the status of the machine tool, and a third area for displaying an action to be taken for the current alarm. Fig. The operation screen shown in Figure 7 further includes an input interface area for inputting whether to accept the new operation screen, and a feedback receiving field 31 for receiving user feedback regarding this operation screen. This feedback receiving field 31 includes a message to assist in operation, asking "Do you accept the new operation screen? (To return to the previous screen, select "Reject")," an "Accept" button, and a "Reject" button. For example, if the user displays the new operation screen and determines that the desired information is included, they can accept this new operation screen by pressing the "Accept" button.For example, if the new operation screen is accepted, the output unit 17 may stop displaying the feedback receiving panel 31 and then display an operation screen including the first area, the second area, and the third area. Otherwise, if the user determines that the desired information is not included in the newly displayed operation screen, or determines that the new operation screen has a layout that displays the desired information in a visually unnoticeable manner, the user may reject the new operation screen by pressing the "Reject" button. If the new operation screen is rejected, the output unit 17 may, for example, stop displaying the new operation screen as shown in FIG. Fig. 7, and then reset the screen to an existing screen display.

[0058] Furthermore, although not shown, the feedback receiving field 31 of the new operation screen may include an interface area that receives not only an input for acceptance or rejection, but also a user comment. Although the above example assumes that the feedback to the new operation screen is received through the alternative "Accept" or "Reject," a change request for a part of the new operation screen may also be received according to the user's operation. For example, in the example shown in Fig. 7, only the information in the third area may be rewritten to indicate an action for the alarm; the display size of each of the areas in the displayed screen may be changed according to the user's operation; and / or the screen display data in the new operation screen may be partially changed. The information received by the feedback receiving field 31 is used as input to the learning device 14.

[0059] As described above, the new operation screen according to the third embodiment includes the feedback receiving panel 31 for receiving user feedback on this new operation screen. Additionally, the feedback receiving panel 31 can receive, as feedback, an operation to return the displayed operation screen from the new operation screen to an existing operation screen. This makes it possible to obtain the user's opinion when the new operation screen generated by the numerical control device 1 does not meet the user's wishes, thereby making the new operation screen closer to an operation screen that better meets the user's requirements. Fourth embodiment.

[0060] Fig. 8 is a diagram showing a functional configuration of a numerical control device 1A according to a fourth embodiment. The numerical control device 1A includes the control unit 11, the operation log acquisition unit 12, the data acquisition unit 13, the learning device 14, the learned model storage unit 15, the inference device 16, the output unit 17, and an operator information storage unit 18. The numerical control device 1A includes the operator information storage unit 18 in addition to the components of the numerical control device 1 according to the first embodiment. The following description mainly explains the differences from the numerical control device 1 according to the first embodiment.

[0061] The operator information storage unit 18 stores operator information for identifying the user operating the numerical control device 1A. The operator information storage unit 18 is capable of storing operator information input via the input device 5 and outputting the operator information to the data acquisition unit 13. The operator information is used as input data in each of the learning device 14 and the inference device 16. This allows the learned model to infer screen display data from the operation log information, the operation state information, and the operator information, and can thus provide an operation screen optimized for each individual user.

[0062] As described above, the numerical control device 1A according to the fourth embodiment further includes the operator information storage unit 18 that stores operator information for identifying the user of the numerical control device 1A, and the learning device 14 and the inference device 16 each use the operator information as input data. Thus, the learned model infers the screen display data from the operator information, the operation log information, and the operation state information. This enables the provision of a user-dependent operation screen. Fifth embodiment.

[0063] A fifth embodiment will be described with reference to an example in which a plurality of numerical control devices 1 distribute a learned model. The first to fourth embodiments were described in which the numerical control device 1 includes the learning device 14 and the inference device 16, and the inference device 16 infers screen display data using a learned model generated in the numerical control devices 1 and 1A in which a process completed in a single machine is executed. In contrast, the fifth embodiment will be described with reference to a method for distributing a learned model by a plurality of numerical control devices 1.

[0064] For example, providing the learning device 14 and the learned model storage unit 15 in a cloud server enables a plurality of numerical control devices 1 to distribute a learned model. Fig. Fig. 9 is a diagram showing an example of a method for distributing a learned model in the fifth embodiment. In this example, providing the learning device 14 in a cloud server or the like allows a plurality of numerical control devices 1-1 to 1-N to communicate with the learning device 14 via a communication channel. For example, the numerical control devices 1-1 to 1-N are each configured not to use any of the components of the Fig. 1 and the learned model storage unit 15 of the numerical control device 1. In this case, the data acquisition unit 13 sends learning data to the learning device 14, and the learned model storage unit 15 sends a learned model to the inference device 16, each of which Fig. 1, via a communication channel such as the Internet.

[0065] It should be noted that the configuration described in this example is merely exemplary, and the method for distributing a learned model by a plurality of numerical control devices 1 in the fifth embodiment is not limited to the above example. For example, the learning device 14 and the learned model storage unit 15 are not limited to those provided in the example in which the learning device 14 and the learned model storage unit 15 are provided in a cloud server, but may be those that can communicate with the plurality of numerical control devices 1-1 to 1-N via a communication channel. Fig. The learning device 14 shown in FIG. 9 can be integrated into a numerical control device 1. Furthermore, the present embodiment is not only applicable to an example of distributing a learned model generated by a single learning device 14. The numerical control devices 1-1 to 1-N may each include the learning device 14 and the learned model storage unit 15 to distribute, via a connection, learned models updated by the respective numerical control devices 1-1 to 1-N.

[0066] As described above, the learning device 14 according to the fifth embodiment is capable of generating a learned model based on the operation log information associated with the plurality of numerical control devices 1-1 to 1-N and the operation state information associated with a plurality of machine tools. Such a learning device 14 can be provided in a cloud server or integrated into a numerical control device 1. Sixth embodiment.

[0067] The numerical control device 1 according to a sixth embodiment has functionality for alerting the user to erroneous operation and potentially risky operation, and for issuing an operation guide based on a result of learning the operation log information of a qualified person when the user is determined to be an inexperienced user based on the operation log information. It should be noted that the numerical control device 1 according to the sixth embodiment has a basic configuration similar to that of the numerical control device 1 of the first embodiment shown in Fig. 1. The following description will therefore primarily describe the differences from the first embodiment.

[0068] Fig. 10 is a diagram for describing functionality of the numerical control device 1 according to the sixth embodiment. The numerical control device 1 according to the sixth embodiment includes a skill level determination unit 19. The skill level determination unit 19 is included, for example, in the data acquisition unit 13. The skill level determination unit 19 determines a user's skill level based on the operation log information and the operation state information, and outputs a determination result to each of the learning devices 14 and the inference device 16. The determination result is used as input data in each of the learning devices 14 and the inference device 16.

[0069] The determination result to be output by the skill level determination unit 19 may be at least one piece of information that can indicate whether the user in question is a novice user. For example, the determination result may be information including a flag indicating whether the user in question is a novice or an experienced user, or information including a numerical value representing a skill level of the user in question. The skill level determination unit 19 is capable of determining that a user is a novice user based on the information of the operation log and the information of the operation state, for example, based on the time required to obtain the desired information, the time required for operation, the number of operation errors, the number of useless operations, and / or the like.

[0070] The inference device 16 may output screen display data including information depending on the determination result of the skill level determination unit 19. For example, in the case of an inexperienced user, the inference device 16 may output screen display data including warning information and / or operation instructions, ie, information that assists in operation.

[0071] As described above, the numerical control device 1 according to the sixth embodiment further includes the skill level determination unit 19 that determines the skill level of a user operating the device based on the operation log information and the operation state information. This allows the learned model to infer, based on the operation log information, the operation state information, and the determination result, screen display data for displaying a new operation screen that includes information sought by the user and extracted from a plurality of existing operation screens, and that includes information depending on the skill level determination result. Seventh embodiment.

[0072] A seventh embodiment will be described with reference to an example of a screen layout of a new operation screen. Fig. Fig. 11 is a diagram showing an example of a new operation screen displayed by the numerical control device 1A according to the seventh embodiment. Note that the numerical control device 1A according to the seventh embodiment has a basic configuration similar to the numerical control device 1A of the fourth embodiment shown in Fig. 8. Therefore, the Fig. 8, and the following description mainly describes the differences from the fourth embodiment.

[0073] The numerical control device 1A is capable of displaying a new operation screen as shown in Fig. 11, a common display area 32 for displaying common information independent of the user and a user-specific display area 33 for displaying information depending on the operating user. The new screen display data inferred using a learned model may include information indicating which information to display in the common display area 32 and which information to display in the user-specific display area 33. Examples of the information to be displayed in the common display area 32 are the operation mode representing a state of the numerical control device 1A, the processing time, and the time.Additionally, the user-specific display area 33 can display expert-level menu options for experienced users and a simplified menu operation guide for novice users. Additionally, the numerical control device 1A can allow the user to specify a range within which the user wishes to maintain the display layout on the operation screen. This can prevent the user from inadvertently changing the layout.

[0074] As described above, according to the seventh embodiment, the novel operation screen displayed by the numerical control device 1A includes the common display area 32, which is a common display area independent of the user, and the user-specific display area 33, which is a display area for displaying different content for a different user. By including the common display area 32, information required regardless of the user can be displayed in this common display area 32, thereby providing a fixed display area. This can reduce the user's burden when recognizing information on the operation screen. Eighth embodiment.

[0075] The numerical control device 1 of an eighth embodiment is capable of using operation log information related to a user-defined screen as input data, where the user-defined screen is an operation screen created by the user. The numerical control device 1 according to the eighth embodiment has a basic configuration similar to the numerical control device 1 of the first embodiment shown in Fig. 1. Therefore, the Fig. 1, and the following description mainly describes the differences from the first embodiment.

[0076] The operation log acquisition unit 12 is capable of obtaining operation log information related to a user-defined screen. Learning also using a user-designed custom screen that is frequently used by the user enables the provision of operation support in a wider variety of use cases. Ninth embodiment.

[0077] A ninth embodiment will be described with respect to a functionality that can propose an optional functionality when the numerical control device 1 includes this optional functionality, which is a functionality that is not available in the standard state but can be activated by purchasing this functionality. The numerical control device 1 according to the ninth embodiment has a basic configuration similar to that of the numerical control device 1 of the first embodiment shown in Fig. 1. Therefore, the Fig. 1, and the following description mainly describes the differences from the first embodiment.

[0078] The learned model generated by the learning device 14 can infer screen display data for displaying a new operation screen containing information about a suggestion to purchase an optional feature if it is determined from the operation log information that making a suggestion for the optional feature is beneficial. In this example, the new operation screen may include, for example, information demonstrating a benefit that arises from using the optional feature.For example, if the numerical control device 1 detects from the operation log information that purchasing an optional feature by the user is likely to provide a benefit, such as reducing the number of operations performed, reducing the number of screen transitions, or reducing operation time, the numerical control device 1 may determine that suggesting the optional feature is beneficial. Therefore, the operation log acquisition unit 12 uses information about existing operation screens, including information about disabled functionality, as input data.Then, if the screen display data to be output includes information about a functionality that the user has not yet acquired, the learned model is able to output screen display data that includes information about a proposal to acquire that functionality.

[0079] As described above, according to the ninth embodiment, the learned model can infer screen display data of an operation screen including information about a suggestion to purchase an optional function of the numerical control device 1. The information about a suggestion to purchase an optional function may include information indicating a benefit that comes from using the optional function. This can further improve the user's work efficiency. Tenth embodiment.

[0080] The numerical control device 1 of a tenth embodiment is capable of displaying an operation screen depending on the number of display devices 4 connected to the numerical control device 1. For example, when the numerical control device 1 is enabled to use multiple display devices 4, including a main monitor and a sub-monitor, the numerical control device 1 provides a display using the multiple display devices 4 depending on the situation. It should be noted that the numerical control device 1 according to the tenth embodiment has a basic configuration similar to that of the numerical control device 1 of the first embodiment shown in Fig. 1. Therefore, the Fig. 1, and the following description mainly describes the differences from the first embodiment.

[0081] The data acquisition unit 13 is capable of receiving, as input data, information representing the number of display devices 4 connected to the numerical control device 1. In this regard, when the numerical control device 1 includes at least one of the display devices 4, "the display devices 4 connected to the numerical control device 1" includes at least one of the display devices 4 included in the numerical control device 1. Further, each of the "display devices 4 connected to the numerical control device 1" may be the display device 4 provided for the numerical control device 1 or a portable device such as a tablet terminal or a smartphone. The learned model infers screen display data from the number of display devices 4, as well as from the operation log information and the operation state information.

[0082] As described above, the learned model according to the tenth embodiment infers screen display data depending on the number of display devices 4 connected to the numerical control device 1. This enables effective use of the display devices 4 available to the numerical control device 1 and can thus further improve the user's work efficiency. Eleventh embodiment.

[0083] An eleventh embodiment will be described with respect to an example of using external data obtained via the Internet or the like. Fig. 12 is a diagram showing a functional configuration of a numerical control device 1B according to the eleventh embodiment. The numerical control device 1B includes the control unit 11, the operation log acquisition unit 12, a data acquisition unit 13B, the learning device 14, the learned model storage unit 15, the inference device 16, and the output unit 17.

[0084] The numerical control device 1B differs from the numerical control device 1 according to the first embodiment in the data obtained by the data acquisition unit 13B. In the following description, the differences from the first embodiment will be mainly described. The data acquisition unit 13B also obtains power information, that is, information representing the power supply of elements such as the numerical control device 1B, the drive unit 2 of the machine tool, and the peripheral device 3, as well as the power consumption in the machine tool. The power information obtained by the data acquisition unit 13B includes, for example, the power consumption of the machine tool obtained from the control unit 11 and power supply information included in external data obtained from outside the numerical control device 1B, for example, via the Internet.The external data includes, for example, information representing an electricity price, the availability of renewable energy, the weather, and / or the like. Such electricity information is used as input data in each of the learning devices 14 and the inference device 16. This allows the learned model to infer screen display data depending on the electricity information. The learning device 14 calculates the reward "r" based on the electricity information and generates a learned model capable of inferring screen display data that enables a reduction in electricity consumption and / or electricity costs.

[0085] In particular, the new screen display data enables the display of a new operation screen which, in addition to the processing time, includes at least environmentally related information such as electricity costs, a greenhouse gas emission volume, a fee for the use of regenerated energy, a suggested processing plan to avoid peak load periods and / or a message to the user to save energy.

[0086] As described above, the learned model according to the eleventh embodiment can infer screen display data that depends on performance information representing the power supply and power consumption of the machine tool. This can assist in selecting appropriate actions via the operation screen with respect to an environmental goal where intuitive selection of actions is difficult due to a complex combination of factors. For example, prioritizing only reducing machining time may have negative effects on the user's achievement of an environmental goal, such as a high electricity bill due to peak demand periods and / or the unavailability of solar power, i.e., power generation from renewable energy, due to bad weather. In contrast, inferring screen data depending on performance information can assist in achieving an environmental goal.

[0087] Next, a hardware configuration of the numerical control devices 1, 1A, and 1B according to the first to eleventh embodiments will be described. The functionality of each applicable component of the numerical control devices 1, 1A, and 1B is implemented using a processing circuit. The processing circuit may be implemented in a dedicated hardware element or a control circuit using a central processing unit (CPU).

[0088] If the above processing circuit is implemented in a dedicated hardware element, such processing circuit is implemented using the Fig. 13 shown processing circuit 90 is implemented. Fig. 13 is a diagram showing a dedicated hardware element for implementing the functionality of the numerical control devices 1, 1A, and 1B according to the first to eleventh embodiments. The processing circuit 90 is a single circuit, a set of multiple circuits, a programmed processor, a parallel programmed processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination thereof.

[0089] When the foregoing processing circuit is implemented in a control circuit using a CPU, this control circuit is, for example, a control circuit 91 as shown in Fig. 14 is configured. Fig. 14 is a diagram showing a configuration of the control circuit 91 for implementing the functionality of the numerical control devices 1, 1A, and 1B according to the first to eleventh embodiments. As shown in Fig. As shown in Figure 14, the control circuit 91 includes a processor 92 and a memory 93. The processor 92 is a CPU, also known as a processing unit, calculation unit, microprocessor, microcomputer, digital signal processor (DSP), and the like. The memory 93 is, for example, a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read-Only Memory), a flash memory, an EPROM (Erasable Programmable ROM), or an electrically erasable programmable ROM (EEPROM) (registered trademark); a magnetic disk, a flexible disk, an optical disk, a compact disc, a MiniDisc, a digital versatile disc (DVD), or the like.

[0090] When the foregoing processing circuit is implemented in the control circuit 91, the processing circuit is implemented such that the processor 92 reads and executes a program corresponding to the processing of each component and stored in the memory 93. The memory 93 is also used as temporary storage during the processing performed by the processor 92. The program can be stored on a storage medium or provided via a communication channel such as the Internet.

[0091] The configurations described in the preceding embodiments are merely examples. These configurations can be combined with other known technologies, and configurations of different embodiments can be combined with each other. Furthermore, some of such configurations may be omitted and / or modified without departing from the spirit thereof. For example, in the preceding embodiments, a technology described as different from the first embodiment can be combined with a functionality of the numerical control device 1A according to the fourth embodiment or with a functionality of the numerical control device 1B according to the eleventh embodiment. List of reference symbols 1, 1A, 1B, 1-1 to 1-N numerical control device; 2 drive unit; 3 peripheral device; 4 display device; 5 input device; 11 control unit; 12 Operation log recording unit; 13, 13B data acquisition unit; 14 learning device; 15 Storage unit for learned models; 16 Inference device; 17 Output unit; 18 Operator information storage unit; 19 Skill level determination unit; 31 feedback receiving field; 32 common display area; 33 user-specific display area; 90 processing circuit; 91 control circuit; 92 processor; 93 storage; 141 learning data acquisition unit; 142 model generation unit; 143 Reward calculation unit; 144 Function update unit; 161 Inference data acquisition unit; 162 Inference unit. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] JP 6898479

[0003] < / nutzungsphase> < / lernphase>

Claims

[1] Numerical control device comprising: a learning data acquisition unit for obtaining learning data including operation log information and operation state information, wherein the operation log information represents a user's operation performed on an operation screen of the numerical control device that controls a machine tool, and also represents information of the operation screen used in the operation, wherein the operation state information represents a state of the machine tool at a time when the operation shown in the operation log information is performed; and a model generation unit for generating a learned model for inferring screen display data for displaying a new operation screen from the operation log information and the operation state information using the learning data, wherein the new operating screen comprises information extracted from several of the existing operating screens. [2] Numerical control device comprising: an inference data acquisition unit for obtaining inference data including operation log information and operation status information, wherein the operation log information represents a user's operation performed on an operation screen of the numerical control device that controls a machine tool, and also represents information of the operation screen used in the operation, wherein the operation state information represents a state of the machine tool at a time when the operation shown in the operation log information is performed; an inference unit for inferring, using a learned model, screen display data for displaying a new operation screen from the operation log information and the operation state information obtained by the inference data acquisition unit, wherein the learned model is a learned model for inferring the screen display data for displaying the new operation screen from the operation log information and the operation state information, wherein the new operating screen comprises information extracted from several of the existing operating screens; and an output unit for causing a display device to output the new operation screen based on the screen display data. [3] The numerical control device according to claim 1 or 2, wherein the new operation screen includes information for reducing an operation burden of the user. [4] The numerical control device according to claim 1 or 2, wherein the new operation screen includes information having a positive effect on an operation result of the user. [5] A numerical control device according to claim 1 or 2, wherein the new operation screen includes a plurality of information separately included in a plurality of existing operation screens. [6] The numerical control device according to any one of claims 1 to 5, wherein the operation log information represents a history of an operation performed by the user on a plurality of pre-existing operation screens. [7] Numerical control device according to one of claims 1 to 6, wherein the plurality of existing operating screens are arranged hierarchically, and wherein the operation log information represents a history of an operation performed by the user on the plurality of hierarchically arranged operation screens already existing. [8] The numerical control device according to any one of claims 1 to 7, wherein the learned model infers the screen display data to display the new operation screen that reduces a number of operation performances, a number of screen transitions, and / or an operation time. [9] The numerical control device according to any one of claims 1 to 8, wherein the screen display data includes at least one kind of information consisting of a design of the new operation screen, a timing of information output, an information output method, or an operation guide, which is information for assisting the user in an operation. [10] The numerical control device according to any one of claims 1 to 9, wherein the operation log information includes a history of time series of coordinate positions input by the user. [11] The numerical control device according to any one of claims 1 to 10, wherein the learned model infers the screen display data to display the new operation screen designed to reduce the number of operations of scrolling, zooming in and / or reducing. [12] The numerical control device according to any one of claims 1 to 11, wherein the new operation screen includes a feedback receiving field for receiving feedback on the new operation screen from the user. [13] The numerical control device according to claim 12, wherein the feedback receiving field receives, as the feedback, a returning operation of a displayed operation screen from the new operation screen to an existing operation screen. [14] Numerical control device according to one of claims 1 to 13, further comprising: an operation information storage unit for storing operation information for identifying the user, where the learned model infers the screen display data from the operation log information, the operation state information and the operation information. [15] The numerical control device according to claim 1, wherein the model generation unit generates the learned model based on the operation log information associated with a plurality of numerical control devices and the operation state information associated with a plurality of machine tools. [16] Numerical control device according to one of claims 1 to 15, further comprising: a skill level determination unit for determining a skill level of the user based on the operation log information and the operation state information, wherein the learned model infers the screen display data for displaying a new operation screen from the operation log information, the operation state information and a determination result of the skill level, wherein the new operation screen includes information searched by the user and extracted from a plurality of existing operation screens, and includes information depending on the determination result. [17] A numerical control device according to any one of claims 1 to 16, wherein the new operation screen includes a common display area independent of the user and a display area for displaying user-specific information. [18] Numerical control device according to one of claims 1 to 17, wherein the operation log information includes the operation log information related to a user-defined screen, where the custom screen is a screen created by the user. [19] Numerical control device according to claim 1 or 2, where the learned model enables the inference of the screen display data of the operating screen, wherein the operation screen includes information of a proposal to purchase an optional functionality of the numerical control device. [20] The numerical control device according to claim 19, wherein the information of a proposal to purchase an optional functionality includes information showing an advantage provided by the optional functionality by using the optional functionality. [21] A numerical control device according to claim 1 or 2, wherein the learned model infers the screen display data depending on a number of display devices associated with the numerical control device. [22] Numerical control device according to claim 1 or 2, where the learned model infers the screen display data depending on performance information, where the power information represents a power supply and a power consumption of the machine tool. [23] Learning device comprising: a learning data acquisition unit for acquiring learning data including operation log information and operation state information, wherein the operation log information represents a user's operation performed on an operation screen of a numerical control device that controls a machine tool, and also represents information of the operation screen used in the operation, wherein the operation state information represents a state of the machine tool at a time when the operation represented in the operation log information is performed; and a model generation unit for generating a learned model for inferring screen display data for displaying a new operation screen from the operation log information and the operation state information using the learning data, wherein the new operating screen comprises information extracted from several of the existing operating screens. [24] Inference device comprising: an inference data acquisition unit for obtaining inference data including operation log information and operation status information, wherein the operation log information represents a user's operation performed on an operation screen of a numerical control device that controls a machine tool, and also represents information of the operation screen used in the operation, wherein the operation state information represents a state of the machine tool at a time when the operation shown in the operation log information is performed; and an inference unit for inferring, using a learned model, screen display data for displaying a new operation screen from the operation log information and the operation state information obtained by the inference data acquisition unit, wherein the learned model is a learned model for inferring the screen display data for displaying the new operation screen from the operation log information and the operation state information, wherein the new operating screen comprises information extracted from several of the existing operating screens. [25] Operation screen display method for a numerical control device that controls a machine tool, wherein the operation screen display method comprises: a step of obtaining inference data including operation log information and operation state information, wherein the operation log information represents a user's operation performed on an operation screen of the numerical control device, and also represents information of the operation screen used in the operation, wherein the operation state information represents a state of the machine tool at a time when the operation shown in the operation log information is performed; a step of inferring screen display data for displaying a new operation screen using a learned model from the operation log information and the operation state information obtained in the step of obtaining the inference data, wherein the learned model is a learned model for inferring the screen display data for displaying the new operation screen from the operation log information and the operation state information, wherein the new operating screen comprises information extracted from several of the existing operating screens; and a step of causing a display device to output the new operation screen based on the screen display data.

Images