Machine learning device, acceleration / deceleration adjustment device, and computer-readable storage medium

Abstract

A machine learning device that estimates parameters that relate to the control of the movement amount in each control cycle and include Nth-order time differential elements (where N is a natural number) of each axis of a machine tool that machines workpieces, the machine learning device comprising: a state observation unit that observes data indicating the movement state of the machine tool; a determination condition acquisition unit that acquires, as determination data, target values relating to the data observed by the state observation unit; a reward calculation unit that calculates, on the basis of the state data and the determination data, rewards associated with machining carried out on the basis of the parameters; a value function unit that updates value functions on the basis of the rewards; and a decision-making unit that estimates, on the basis of the value functions, combinations of set values of the parameters that are more suitable for the machining, and outputs said combinations.

Description

[Technical field]

[0001] The present invention relates to a machine learning device, an acceleration / deceleration adjustment device, and a computer-readable storage medium. [Background technology]

[0002] Machine tools are controlled based on a machining program to machine workpieces and manufacture products such as parts and molds. The machining speed when machining a workpiece is commanded as the axis movement speed in the machining program. The axis movement speed commanded in the machining program is the maximum speed of relative movement between the tool and workpiece (tool movement). The machine tool varies the axis movement speed according to parameters related to the control of each axis at the start of machining, around corners, curved sections, etc., within a range that does not exceed the commanded maximum speed.

[0003] In the manufacture of products, the target tolerances and machining surface quality are set in advance. The target machining time is also determined in advance. The machine tool operator checks the machining error and machining surface quality of the machined product, while adjusting parameters such as acceleration and deceleration time constants, and adjusting the movement speed commanded in the machining program.

[0004] As a conventional technology for adjusting the parameters related to the control of each axis in product machining, a patent has been applied for that uses machine learning technology to obtain an optimal speed distribution that balances machining error, machined surface quality, and machining time (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2017-068325 A Summary of the Invention [Problem to be solved by the invention]

[0006] When using data related to the velocity distribution as a criterion for parameter adjustment, it is necessary to specify threshold values ​​for the acceleration, jerk, etc. that serve as the criterion. However, unless it is understood how to set the threshold values ​​and what level of machining error and machined surface quality will result, it is difficult to quantitatively control the target tolerance and machined surface quality.

[0007] In addition, since the appropriate speed distribution is set in accordance with a specific machining purpose for a specific machine tool, there is a problem in that the appropriate speed distribution must be reset every time the machine tool performing machining is changed or the machining purpose is changed. Therefore, there is a need for a technique that allows adjustment of processing parameters based on criteria other than the velocity distribution. [Means for solving the problem]

[0008] The acceleration / deceleration adjustment device according to the present disclosure solves the above problems by directly specifying and quantitatively controlling machining errors such as allowable shape errors and position deviations, and machining surface quality. The acceleration / deceleration adjustment device according to the present disclosure introduces machine learning to optimize a combination of parameter settings that control the movement amount of each axis for each control period, including an N-th order time differential element (N is a natural number) of the speed of each axis.

[0009] One aspect of the present disclosure is a machine learning device that estimates parameters related to control of a movement amount for each control period including an N-th order time differential element (N is a natural number) of each axis of a machine tool that machines a workpiece, the machine learning device including: a state observation unit that observes information related to at least one of machining accuracy and machined surface quality in the machining, and a machining time required for the machining, as data indicating an operating state of the machine tool; a judgment condition acquisition unit that acquires, as judgment data, a target value related to the data observed by the state observation unit; a reward calculation unit that calculates a reward for machining based on the parameters based on the data observed by the state observation unit and the judgment data acquired by the judgment condition acquisition unit; and a value function that updates a value function for calculating a value of a machining state based on the parameters based on the reward. Update departmentand a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the processing based on the updated value function, and outputs the estimated combination of setting values ​​of the parameters.

[0010] Another aspect of the present disclosure is an acceleration / deceleration adjustment device that adjusts parameters related to control of the movement amount per control period including N-th order time differential elements (N is a natural number) of each axis of a machine tool that processes a workpiece, and includes: a state observation unit that observes information related to at least one of the machining accuracy and the machined surface quality in the machining, and the machining time required for the machining, as data indicating the operating state of the machine tool; a judgment condition acquisition unit that acquires a target value related to the data observed by the state observation unit as judgment data; a value function memory unit that stores a value function for calculating a value of the machining state based on the parameters; a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the machining based on the value function, and outputs the estimated combination of setting values ​​of the parameters; and an action output unit that adjusts the parameters of the machine tool based on the combination of setting values ​​of the parameters output by the decision-making unit.

[0011] Another aspect of the present disclosure is a computer-readable storage medium storing a program for operating a computer as a machine learning device that estimates parameters related to control of a movement amount per control period including an N-th order time differential element (N is a natural number) of each axis of a machine tool that processes a workpiece, the machine learning device including a state observation unit that observes information related to at least one of machining accuracy and machined surface quality in the machining and a machining time required for the machining as data indicating an operating state of the machine tool, a judgment condition acquisition unit that acquires a target value related to the data observed by the state observation unit as judgment data, a reward calculation unit that calculates a reward for machining based on the parameters based on the data observed by the state observation unit and the judgment data acquired by the judgment condition acquisition unit, and a value function that updates a value function for calculating a value of a machining state based on the parameters based on the reward. Update departmentand a computer-readable storage medium storing a program for causing a computer to operate as a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the processing based on the updated value function, and outputs the estimated combination of setting values ​​of the parameters.

[0012] Another aspect of the present disclosure is a computer-readable storage medium storing a program for operating a computer as an acceleration / deceleration adjustment device that adjusts parameters related to control of the movement amount per control period, including N-th order time differential elements (N is a natural number) of each axis of a machine tool that machines a workpiece, comprising: a state observation unit that observes information related to at least one of the machining accuracy and the machined surface quality in the machining, and the machining time required for the machining, as data indicating the operating state of the machine tool; a judgment condition acquisition unit that acquires a target value related to the data observed by the state observation unit as judgment data; a value function memory unit that stores a value function for calculating a value of the machining state based on the parameters; a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the machining based on the value function, and outputs the estimated combination of setting values ​​of the parameters; and a behavior output unit that adjusts the parameters of the machine tool based on the combination of setting values ​​of the parameters output by the decision-making unit. Effect of the Invention

[0013] According to one aspect of the present disclosure, by setting target values ​​(shape error, position deviation, etc.) for the actual machining accuracy and machined surface quality, it becomes possible to adjust the parameters to be more suitable, thereby enabling quantitative control of the machining accuracy and machined surface quality. [Brief description of the drawings]

[0014] [Figure 1] FIG. 2 is a hardware configuration diagram of the acceleration / deceleration adjustment device. [Diagram 2] 2 is a block diagram showing the functions of the acceleration / deceleration adjustment device; FIG. [Diagram 3] FIG. 11 is a diagram illustrating calculation of processing accuracy. [Figure 4] FIG. 13 is a diagram for explaining calculation of the processed surface quality. [Diagram 5] 4 is a flowchart showing a schematic operation example of the acceleration / deceleration adjustment device. [Figure 6] FIG. 11 is a diagram illustrating parameter adjustment. [Figure 7] FIG. 11 is a block diagram showing the functions of an acceleration / deceleration adjustment device according to another embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a schematic hardware configuration diagram showing the main parts of an acceleration / deceleration adjustment device according to one embodiment of the present invention. The acceleration / deceleration adjustment device 1 of the present invention can be implemented as, for example, a control device that controls a machine tool based on a machining program. The acceleration / deceleration adjustment device 1 of the present invention can also be implemented on a computer such as a personal computer attached to a control device that controls a machine tool based on a machining program, or a personal computer, cell computer, fog computer 6, cloud server 7, or the like that is connected to the control device via a wired / wireless network. In this embodiment, the acceleration / deceleration adjustment device 1 is implemented as a personal computer connected to a control device that controls a machine tool based on a machining program, or the like that is connected to a control device via a wired / wireless network. of This shows an example of implementation on a PC connected to a control device that controls a machine tool via the

[0016] The CPU 11 provided in the acceleration / deceleration adjustment device 1 according to this embodiment is a processor that controls the entire acceleration / deceleration adjustment device 1. The CPU 11 reads out a system program stored in the ROM 12 via the bus 22, and controls the entire acceleration / deceleration adjustment device 1 in accordance with the system program. The RAM 13 temporarily stores temporary calculation data, display data, various data input from outside, and the like.

[0017] The non-volatile memory 14 is composed of, for example, a memory backed up by a battery (not shown) or an SSD (Solid State Drive), and retains its stored state even when the power to the acceleration / deceleration adjustment device 1 is turned off. The non-volatile memory 14 stores data read from an external device 72 via an interface 15, data input via an input device 71, data acquired from the machine tool 3 (including data detected by the sensor 4), and the like. The data stored in the non-volatile memory 14 may be expanded in the RAM 13 at the time of execution / use. In addition, various system programs such as known analysis programs are written in advance in the ROM 12.

[0018] Machine tool 3 is fitted with sensors 4 that detect physical quantities such as current, voltage, vibration, etc. of each part when machine tool 3 is operating. Examples of machine tool 3 include a machining center and a lathe. In response to a request from acceleration / deceleration adjustment device 1, machine tool 3 transmits data such as the position, speed, acceleration, jerk, vibration, and machining time of each axis during machining via network 5.

[0019] The interface 15 is an interface for connecting the CPU 11 of the acceleration / deceleration adjustment device 1 to an external device 72 such as a USB device. For example, pre-stored machining programs and data related to the operation of each machine tool 3 can be read from the external device 72. In addition, the machining programs and setting data edited within the acceleration / deceleration adjustment device 1 can be stored in an external storage means via the external device 72.

[0020] The interface 20 is an interface for connecting the CPU of the acceleration / deceleration adjustment device 1 to a wired or wireless network 5. The network 5 is connected to the machine tool 3, fog computer 6, cloud server 7, etc., and exchanges data with the acceleration / deceleration adjustment device 1.

[0021] The display device 70 displays various data loaded into the memory, data obtained as a result of executing a program, data output from the machine learning device 100 described below, and the like, which are output via an interface 17. An input device 71 consisting of a keyboard, a pointing device, and the like passes commands and data based on operations by an operator to the CPU 11 via an interface 18.

[0022] The interface 21 is an interface for connecting the CPU 11 and 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 that stores a system program and the like, a RAM 103 that performs temporary storage in each process related to machine learning, and a non-volatile memory 104 that is used to store models and the like. The machine learning device 100 can observe each piece of information that can be acquired by the acceleration / deceleration adjustment device 1 via the interface 21 (for example, data detected during machining of the machine tool 3). In addition, the acceleration / deceleration adjustment device 1 acquires the processing results output from the machine learning device 100 via the interface 21, and stores or displays the acquired results, or transmits them to another device via a network 5 or the like.

[0023] 2 is a schematic block diagram showing functions of the acceleration / deceleration adjustment device 1 according to the first embodiment of the present invention. Each function of the acceleration / deceleration adjustment device 1 according to this embodiment is realized by the CPU 11 of the acceleration / deceleration adjustment device 1 shown in FIG. 1 and the processor 101 of the machine learning device 100 executing a system program and controlling the operation of each part of the acceleration / deceleration adjustment device 1 and the machine learning device 100.

[0024] The acceleration / deceleration adjustment device 1 of this embodiment includes a state observation unit 110, a judgment condition acquisition unit 120, and a behavior output unit 150. The machine learning device 100 of the acceleration / deceleration adjustment device 1 includes a learning unit 130 and a decision-making unit 140. The RAM 103 to the non-volatile memory 104 of the machine learning device 100 further include a learning unit 130 and a decision-making unit 140. 130A value function storage unit 138 is prepared in advance to store a value function as a result of machine learning by the above-mentioned method.

[0025] The state observing unit 110 observes information related to at least one of the machining accuracy and the quality of the machined surface, and the machining time as data indicating the operating state of the machine tool. Observing here means acquiring data from the environment and calculating predetermined data based on the acquired data. First, the state observing unit 110 acquires various data detected during the operation of the machine tool 3 as data indicating the operating state of machining by the machine tool 3. The state observing unit 110 acquires, for example, the position, speed, acceleration, jerk, vibration, machining time, etc. of each axis during machining in the machine tool 3 as data indicating the operating state of machining by the machine tool 3. In addition, the state observing unit 110 acquires parameters (linear acceleration, linear jerk, post-interpolation acceleration / deceleration time constant, corner speed difference, position loop gain, feedforward coefficient, etc.) related to the control of the movement amount for each control cycle including the N-th order time differential element (N is a natural number) of each axis set during machining by the machine tool 3, and the machining program used for controlling machining, as data indicating the operating state of machining by the machine tool 3. The data acquired by the state observing unit 110 may be instantaneous values ​​acquired at a predetermined timing, or may be time-series data acquired over a predetermined period of time.

[0026] Furthermore, the state observing unit 110 calculates data relating to the machining accuracy and the machined surface quality in the machining based on the position data, speed data, acceleration data, jerk data, vibration data, etc. of each axis included in the data indicating the operational state of machining by the machine tool 3. Examples of the calculated data relating to the machining accuracy and the machined surface quality include shape error, position deviation, vibration error, etc.

[0027] An example of calculation of machining accuracy (shape error) and machining surface quality (position deviation) by the state observation unit 110 will be described with reference to FIG. 3. FIG. 3 shows a machining program path commanded by a machining program executed in the machine tool 3 and a movement path calculated based on the position of the motor. In FIG. 3, the horizontal axis indicates the X-coordinate position, and the vertical axis indicates the Y-coordinate position. The solid line arrow indicates the machining program path commanded by the machining program, and the dotted line arrow indicates the movement path calculated based on the position of the motor. In machining by the machine tool 3, a movement command to be output to the motor of each axis along the machining program path commanded by the machining program is calculated based on parameters related to control of the movement amount per control period including the N-th order time differential element (N is a natural number) of each axis. In this calculation, a movement command is calculated that prioritizes the machining speed, efficiency, etc. within the range of the set parameters. Therefore, when the motor is moved based on the calculated movement path, the movement path calculated based on the position of the motor (the relative movement path between the tool driven by the motor and the workpiece) does not match the machining program path. The state observing unit 110 calculates a shape error indicating the machining accuracy and a position deviation indicating the quality of the machined surface based on the difference between the machining program path and a movement path calculated based on the motor position. For the shape error, the maximum value Emax of the shape error may be used, for example. For the position deviation, the average value Emean, the variance value Edist, etc. may be used.

[0028] An example of calculation of the machined surface quality (vibration error) by the state observing unit 110 will be described with reference to FIG. 4. FIG. 4 shows a machining program path commanded by a machining program executed in the machine tool 3 and a movement path (with / without vibration) calculated based on the position of the motor. Looking more closely at the path calculated based on the motor position, the motor position data vibrates. This vibration appears as scratches and streaks during machining, and affects the machined surface quality. The state observing unit 110 calculates a vibration error indicating the machined surface quality based on the difference between the movement command output to the motor and the position acquired from the motor. For the vibration error, for example, the maximum amplitude Amax or the average amplitude Amean may be used.

[0029] If a vibration measuring device can be prepared as the sensor 4, the vibration error may be calculated using vibration data from the vibration measuring device in addition to the motor position data. The vibration data from the vibration measuring device can obtain vibration closer to the machining point (contact position between the tool and the workpiece).

[0030] The state observing unit 110 may acquire data directly from the machine tool 3 via the network 5. The state observing unit 110 may acquire data acquired and stored by an external device 72, the fog computer 6, the cloud server 7, etc. The data acquired or calculated by the state observing unit 110 is input to the learning unit 130 and the decision making unit 140.

[0031] The judgment condition acquisition unit 120 acquires judgment data related to the machining purpose in machining by the machine tool 3. The judgment data related to the machining purpose includes, for example, tolerance values ​​related to machining accuracy and machining surface quality, such as a predetermined allowable machining accuracy (allowable shape error), allowable machining surface quality (allowable position deviation), and allowable machining surface quality (allowable vibration error). The judgment data related to the machining purpose includes, for example, a target machining time. The judgment condition acquisition unit 120 may acquire tolerance values ​​related to machining accuracy and machining surface quality set in the machine tool 3 via the network 5. The judgment condition acquisition unit 120 may acquire data stored in the external device 72, the fog computer 6, the cloud server 7, and the like. The judgment condition acquisition unit 120 may prompt the operator to input tolerance values ​​related to machining accuracy and machining surface quality and a target machining time from the input device 71. The data acquired by the judgment condition acquisition unit 120 is input to the learning unit 130 and the decision-making unit 140.

[0032] Learning unit 130 executes processing related to machine learning based on data indicating the operating state of machining by machine tool 3 acquired by state observation unit 110 and judgment data related to the machining purpose acquired by judgment condition acquisition unit 120. Learning unit 130 includes reward calculation unit 132 and value function update unit 134. Learning unit 130 updates the value function by value function update unit 134 based on the reward calculated by reward calculation unit 132, thereby learning the correlation between a combination of parameters related to control of the movement amount for each control cycle, including an N-th order time differential element (N is a natural number) of each axis, and the value of the combination.

[0033] The remuneration calculation unit 132 calculates a remuneration for the current operating state of the machine tool 3 based on data indicating the operating state of machining by the machine tool 3 and judgment data related to the machining purpose. The remuneration calculation unit 132 compares the values ​​indicating the machining accuracy and the machining surface quality calculated by the state observation unit 110 with the judgment data related to the machining purpose, and calculates a remuneration by a predetermined remuneration calculation formula set in advance based on the comparison result. The judgment data related to the machining purpose includes an allowable value related to the machining accuracy and the machining surface quality. The remuneration calculation unit 132 calculates a high remuneration when the calculated values ​​indicating the machining accuracy and the machining surface quality fall within this allowable value. Moreover, the remuneration calculation unit 132 calculates a low remuneration when the calculated values ​​indicating the machining accuracy and the machining surface quality exceed this allowable value. The remuneration calculation unit 132 may calculate a higher remuneration according to the degree to which the values ​​fall within the allowable value. Moreover, the remuneration calculation unit 132 may calculate a lower remuneration according to the degree to which the values ​​exceed the allowable value. The remuneration calculation unit 132 may calculate a negative remuneration.

[0034] The remuneration calculation unit 132 further compares a value indicating the machining time required for machining in the machine tool 3 with a target machining time included in the judgment data related to the machining purpose, and calculates an additional remuneration using a preset remuneration calculation formula based on the comparison result. The remuneration calculation unit 132 calculates a high remuneration when the machining time required for machining falls within the target machining time. The remuneration calculation unit 132 also calculates a high remuneration when the machining time required for machining falls within the target machining time. AimIf the processing time is exceeded, a lower remuneration is calculated. The remuneration calculation unit 132 may calculate a higher remuneration depending on the degree to which the processing time required for processing is met. The remuneration calculation unit 132 may also calculate a lower remuneration depending on the degree to which the processing time required for processing is exceeded. The remuneration calculation unit 132 may calculate a negative remuneration. The remuneration calculation unit 132 may use the additional remuneration calculated in this manner to calculate the processing accuracy and the processed surface quality. rank This will be added to the compensation calculated based on the above.

[0035] When the machining time is considered as the remuneration, the remuneration calculation unit 132 may store the machining time when the machining accuracy and the machined surface quality are within the allowable values. At this time, the shortest machining time among the stored machining times is set as the judgment data related to the machining purpose. Then, the remuneration calculation unit 132 uses the machining time at that time as the judgment data only when the machining accuracy and the machined surface quality are within the allowable values. and The reward is calculated based on the shortest stored machining time. By doing this, it becomes possible to search for parameters that will shorten the machining time within the range of the target machining accuracy and machined surface quality.

[0036] The value function update unit 134 updates the value function stored in the value function storage unit 138 based on the reward calculated by the reward calculation unit 132. The value function used in the present invention is a state value function that calculates the value of being in a current state by taking a combination of parameters related to the control of the movement amount per control cycle, including an N-order time differential element (N is a natural number) of each axis set during machining of the machine tool 3, as a state. The state value function V according to this embodiment may be defined as a function that returns the reward calculated by the reward calculation unit 132 as a value when machining is performed, for example, by taking each state (a combination of parameters related to the control of the movement amount per control cycle, including an N-order time differential element (N is a natural number) of each axis). It is preferable that the state value function is set to output a high value for all possible states at the stage when learning is started. The state value function may be constructed as a multilayer neural network in which a combination of parameters related to the control of the movement amount per control cycle, including an N-order time differential element (N is a natural number) of each axis set during machining of the machine tool 3, is used as input data, and the value of the state of the combination of parameters is used as output data.

[0037] The decision-making unit 140 outputs a combination of parameters related to the control of the movement amount for each control period, including the N-th order time differential element (N is a natural number) of each axis, based on the value function created by the learning unit 130 through machine learning. The decision-making unit 140 uses the value function to find a combination of parameters with a higher value than the combination of parameters set in the currently performed machining. For example, the decision-making unit 140 compares a value calculated from the currently set combination of parameters with a value calculated from the combination of parameters when each parameter is changed by a predetermined amount. For example, the decision-making unit 140 calculates a value using the value function for each of the currently set combination of parameters, a combination of parameters in which the linear acceleration is changed by +ΔA, a combination of parameters in which the linear acceleration is changed by -ΔA, a combination of parameters in which the post-interpolation acceleration / deceleration time constant is changed by +Δτ, a combination of parameters in which the post-interpolation acceleration / deceleration time constant is changed by -Δτ, ... to find a combination of parameters with a higher value. Then, the determined combination of parameters is output as a combination of parameters more suitable for the current machining. At this time, if the value calculated from the currently set parameter combination is the highest, the decision-making unit 140 may output the current parameter combination as the parameter combination with the higher value. If there are parameter combinations with the same value, the decision-making unit 140 may output a random parameter combination from among the parameter combinations other than the currently set one. In the early stage of learning by the learning unit 130, the decision-making unit 140 may output a random parameter combination with a certain probability (ε-greedy method), regardless of the value calculated by the value function. In this way, a more appropriate parameter combination can be efficiently searched for.

[0038] The behavior output unit 150 judges whether or not to continue the parameter adjustment based on the combination of parameters output by the decision-making unit 140. When it is judged that the parameter adjustment is to be continued, the behavior output unit 150 sets the combination of parameters output by the decision-making unit 140 in the machine tool 3 and commands the machine tool 3 to perform a machining operation again. For example, the behavior output unit 150 may judge to continue the parameter adjustment when the combination of parameters output by the decision-making unit 140 is different from the combination of parameters currently set in the machine tool 3. In addition, the behavior output unit 150 may record the number of times the parameters of the machine tool 3 have been changed since the start of the parameter adjustment, and may judge to continue the parameter adjustment when the number of times the parameters have been changed is within a predetermined number of times.

[0039] The behavior output unit 150 may directly set the parameter combination to the machine tool 3 via the network 5. Moreover, the behavior output unit 150 may transmit the parameter combination to the fog computer 6 or the cloud server 7 via the network 5, and indirectly prompt the machine tool 3 to set the parameters. Furthermore, the behavior output unit 150 may display the parameter combination on the display device 70, and prompt the operator to set the parameters in the machine tool 3.

[0040] 5 is a flow chart illustrating a process for determining a combination of parameters using the acceleration / deceleration adjustment device 1 having the above-mentioned configuration. When an operator or the like issues a command to adjust parameters related to control of the movement amount per control period including an N-th order time differential element (N is a natural number) of each axis, the acceleration / deceleration adjustment device 1 first determines the combination of parameters (for example, linear acceleration of 4000 mm / sec) previously set by the operator or the like. 2 , corner speed difference of 800 mm / min, post-interpolation acceleration / deceleration time constant of 32 msec, etc.) are set for the machine tool 3. Then, the machine tool 3 is instructed to perform an idling run according to a given machining program given in advance (step SA01).

[0041] While the machine tool 3 receives this command and performs an idle operation of the machining program, the state observing unit 110 acquires data indicating the operating state of the machine tool 3 (time series data of the motor position, time series data of the motor speed, machining time, etc.) (step SA02). Then, based on the acquired data indicating the operating state of the machine tool 3, data related to the machining accuracy and machined surface quality (for example, machining accuracy (shape error): 80 μm, machined surface quality (position deviation): 8 μm, machined surface quality (vibration error): 0.09 μm, etc.) are calculated (step SA02). In addition, the judgment condition acquiring unit 120 acquires judgment data related to the machining purpose in machining by the machine tool 3 (machining accuracy (allowable shape error): 100 μm, machined surface quality (allowable position deviation): 10 μm, machined surface quality (allowable vibration error): 0.1 μm, allowable machining time 12.0 sec, etc.) (step SA03).

[0042] Reward calculation unit 132 calculates the reward for the current parameter combination based on the data related to the machining accuracy and the machined surface quality input from state observation unit 110 and the judgment data related to the machining purpose input from judgment condition acquisition unit 120 (step SA04). Then, based on the calculated reward, value function update unit 134 updates the value function stored in value function storage unit 138 (step SA05).

[0043] Based on the updated value function, decision-making unit 140 determines a parameter combination that is considered more appropriate for the current machining, and outputs the determined parameter combination. Action output unit 150, which has received this, determines whether or not to continue parameter adjustment, and if it determines to continue parameter adjustment, sets the parameter combination output by decision-making unit 140 in machine tool 3, and issues a command to perform idle operation again using the machining program with the set parameters (step SA06).

[0044] The acceleration / deceleration adjustment device 1 having the above configuration can quantitatively control parameters suitable for the actual machining accuracy and the machining surface quality by setting target values ​​(shape error, position deviation, etc.). By quantitatively controlling the combination of parameter setting values, it becomes possible to hold a combination of parameter setting values ​​suitable for a specific machining purpose, and it is possible to switch appropriately according to the machining purpose. Furthermore, the optimization of the combination of parameter setting values ​​by the acceleration / deceleration adjustment device 1 according to this embodiment can be performed while observing the environment (control device, machine tool) from the outside. Therefore, there is no need to incorporate new software into the environment, and it can be used in a wide range of environments.

[0045] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described examples of the embodiments, and can be embodied in various forms by making appropriate modifications. For example, in the above embodiment, the evaluation of the machining accuracy and the quality of the machined surface is performed based on a preset remuneration calculation formula, but the evaluation program related to the evaluation may be configured to be registered from outside. By configuring in this way, even if the machining content changes and a parameter to be learned is added, it becomes possible to efficiently evaluate the machining accuracy and the quality of the machined surface by providing a dedicated evaluation program for square corners, R-angle corners, etc.

[0046] The range of each parameter output by decision making unit 140 may be set in advance. By configuring in this way, the search range of the parameters can be limited.

[0047] In the above embodiment, the decision-making unit 140 determines the combination of parameters to be output based on the value output by the value function, but it may output a combination of parameters created by adjusting parameters that are considered to be more appropriate according to a predetermined rule, for example. FIG. 6 shows a table showing the adjustment direction of parameters to solve a problem to be solved that is identified from the processing result, and the priority order of the adjustment direction. In this way, by storing the relationship between the problem to be solved and the parameter adjustment method as a rule, the decision-making unit 140 can determine the parameters to be adjusted by referring to the rule.

[0048] The remuneration related to the machining accuracy and the quality of the machined surface calculated by the remuneration calculation unit 132 and the additional remuneration related to the machining time are weighted. setting By configuring in this way, it becomes possible to perform fine adjustments according to the purpose of machining, such as increasing the weight of the reward related to the machining accuracy and the quality of the machined surface when machining quality is emphasized, and increasing the weight of the reward related to the machining time when machining time is emphasized.

[0049] In the above embodiment, a configuration has been shown in which parameters are adjusted while machine learning is performed by learning unit 130, but once learning by learning unit 130 has been sufficiently performed, learning unit 130 may be deleted from acceleration / deceleration adjustment device 1 while leaving value function storage unit 138. Fig. 7 shows an example of a configuration in which learning unit 130 has been deleted. With such a configuration, the value function is no longer updated, but by providing initial parameters to acceleration / deceleration adjustment device 1, decision-making unit 140 can search for more appropriate parameters and adjust the parameters of machine tool 3. [Explanation of symbols]

[0050] 1 Acceleration / deceleration adjustment device 3 Machine tools 4 Sensors 5. Network 6. Fog Computer 7. Cloud Server 11 CPU 12 ROM 13 RAM 14 Non-volatile memory 15 Interface 17,18,20,21 Interface 22 Bus 70 Display device 71 Input Devices 72 External equipment 100 Machine Learning Device 101 Processor 102 ROM 103 RAM 104 Non-volatile memory R 1 10 Status Observation Section 120 Judgment condition acquisition section 130 Learning Department 132 Remuneration Calculation Department 134 Value function update unit 138 Value Function Memory Unit 140 Decision Making Department 150 Behavioral Output Unit

Claims

1. A machine learning device that estimates parameters related to control of a movement amount for each control period including an N-th order time differential element (N is a natural number) of each axis of a machine tool that processes a workpiece, a status observation unit that observes information relating to at least one of a machining accuracy and a machined surface quality in the machining and a machining time required for the machining as data indicating an operating status of the machine tool; a judgment condition acquisition unit that acquires a target value related to the data observed by the state observation unit as judgment data; a reward calculation unit that calculates a reward for the processing based on the parameters based on the data observed by the state observation unit and the judgment data acquired by the judgment condition acquisition unit; a value function update unit that updates a value function for calculating a value of a processing state based on the parameters based on the remuneration; a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the processing based on the updated value function, and outputs the estimated combination of setting values ​​of the parameters; A machine learning device comprising:

2. An evaluation program capable of evaluating at least one of the machining accuracy and the machining surface quality can be registered, and a remuneration relating to at least one of the machining accuracy and the machining surface quality is calculated using the evaluation program. The machine learning device according to claim 1 .

3. the decision-making unit estimates a combination of setting values ​​of the parameters that is more suitable for the machining and that has a shorter machining time based on the updated value function, and outputs the estimated combination of setting values ​​of the parameters. The machine learning device according to claim 1 .

4. An acceleration / deceleration adjustment device that adjusts parameters related to control of a movement amount per control period including an N-th order time differential element (N is a natural number) of each axis of a machine tool that processes a workpiece, a status observation unit that observes information relating to at least one of a machining accuracy and a machined surface quality in the machining and a machining time required for the machining as data indicating an operating status of the machine tool; a judgment condition acquisition unit that acquires a target value related to the data observed by the state observation unit as judgment data; A value function storage unit that stores a value function for calculating a value of a machining state based on the parameters; a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the processing based on the value function, and outputs the estimated combination of setting values ​​of the parameters; a behavior output unit that adjusts the parameters of the machine tool based on the combination of setting values ​​of the parameters output by the decision making unit; An acceleration / deceleration adjustment device comprising:

5. The judgment condition acquisition unit is capable of setting a plurality of target values. The acceleration / deceleration adjusting device according to claim 4.

6. The range of the set value of the parameter output by the decision-making unit can be set. The acceleration / deceleration adjusting device according to claim 4.

7. The number of times the parameters are adjusted can be set. The acceleration / deceleration adjusting device according to claim 4.

8. A computer-readable storage medium storing a program for operating a computer as a machine learning device that estimates parameters related to control of a movement amount for each control period including an N-th order time differential element (N is a natural number) of each axis of a machine tool that processes a workpiece, a status observation unit that observes information relating to at least one of a machining accuracy and a machined surface quality in the machining and a machining time required for the machining as data indicating an operating status of the machine tool; a judgment condition acquisition unit that acquires a target value related to the data observed by the state observation unit as judgment data; a reward calculation unit that calculates a reward for the processing based on the parameters based on the data observed by the state observation unit and the judgment data acquired by the judgment condition acquisition unit; a value function update unit that updates a value function for calculating a value of a processing state based on the parameters based on the remuneration; a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the processing based on the updated value function, and outputs the estimated combination of setting values ​​of the parameters; A computer-readable storage medium storing a program for operating a computer.

9. A computer-readable storage medium storing a program for operating a computer as an acceleration / deceleration adjustment device that adjusts parameters related to control of a movement amount per control period including an N-th order time differential element (N is a natural number) of each axis of a machine tool that processes a workpiece, a status observation unit that observes information relating to at least one of a machining accuracy and a machined surface quality in the machining and a machining time required for the machining as data indicating an operating status of the machine tool; a judgment condition acquisition unit that acquires a target value related to the data observed by the state observation unit as judgment data; A value function storage unit that stores a value function for calculating a value of a machining state based on the parameters; a decision-making unit that estimates a combination of setting values ​​of the parameters that is more suitable for the processing based on the value function, and outputs the estimated combination of setting values ​​of the parameters; a behavior output unit that adjusts the parameters of the machine tool based on the combination of setting values ​​of the parameters output by the decision-making unit; A computer-readable storage medium storing a program for operating a computer.

Images