Control unit for a wire EDM machine and control method for a wire EDM machine

Abstract

Control unit (20, 30) for a wire EDM machine (1; 1a; 1b; 1c) which machine-machines a workpiece (11) by applying a voltage between the workpiece (11) and an electrode (10) and causing an electrical discharge, wherein the control unit (20, 30) is configured to control a drive trajectory of the electrode (10) relative to the workpiece (11), a relative machine-machining speed between the workpiece (11) and the electrode (10), and a frequency of electrical discharges of a voltage periodically applied between the electrode (10) and the workpiece (11), wherein the control unit (20, 30) comprises: a storage device (36, 37, 38, 39) which stores at least one element of data selected from a voltage mean value, which is a mean value of a voltage applied between the workpiece (11) and the electrode (10), the frequency of electrical discharges, the machine processing speed and the drive trajectory of an (n-1)th machine operation of n machine operations of a predetermined machine processing section of the workpiece (11), where n is an integer of two or more; and a calculating device (342) which calculates an electrode gap, which is a distance between the workpiece (11) and the electrode (10), of an nth machining operation using a calculation model which specifies a relationship between the data of the (n-1)th machining operation and a machining shape of the workpiece (11), and which calculates from the electrode gap of the nth machining operation at least one command value, which is selected from the machining speed, the frequency of electrical discharges and the drive trajectory of the nth machining operation, which corresponds to a machining quantity required to achieve a desired shape, which is based on a machining program for machining the machining section.

Description

Area

[0001] The present disclosure relates to a control device for a wire EDM machine and a control method for a wire EDM machine for controlling a wire EDM machine which machine-machines a workpiece by applying a voltage between the workpiece and an electrode and causing an electrical discharge. background

[0002] An electrical discharge machining (EDM) machine is a device that machines a workpiece by generating an arc discharge between electrodes, that is, between the machining electrode and the workpiece. The EDM machine requires a power source to generate this electrical discharge. By applying a high voltage between the electrodes or by reducing the distance between them to increase the electric field strength, an electrical discharge is generated due to dielectric breakdown, thus performing the material removal process on the workpiece.To induce another electrical discharge after the completion of an electrical discharge and dielectric recovery, it is necessary, because the distance between the electrodes is large, to apply a high voltage between the electrodes or to narrow the space between the electrodes to increase the electric field intensity. Machining with the electrical discharge machining (EDM) machine is repeated several times under varying machining conditions, which depend on the target parameters and the required surface roughness accuracy. Specifically, as a process to produce the target shape from the workpiece, a rough machining operation is first performed, followed by a shape correction machining operation to improve the shape accuracy and reduce the surface roughness according to the target shape.

[0003] In shape correction machining, it is necessary to improve the accuracy of the surface roughness and correct the shape deviations introduced in previous machining operations, including rough machining. The direction and extent of the shape deviation relative to the machining progress direction in these previous operations vary depending on the shape of the workpiece, the machining progress direction, the machining conditions, and other factors.

[0004] Shape correction machining requires the ability to perform machining according to the target parameters regardless of changes in the amount of machining required, which depend on the machining location and direction relative to the machining progress. If the shape cannot be corrected according to the target parameters in shape correction machining, the electrode spacing (also referred to here as the "inter-electrode distance"), which is the distance between the electrodes, varies depending on the machining location, leading to a high probability of increased variations in surface roughness and machining parameters.Therefore, patent literature 1 discloses a technique for monitoring the inter-electrode mean machine machining voltage to detect the machine machining state, including the frequency of electrical discharges and the amount of machine machining during a spark EDM machine machining operation, and for controlling the relative movement speed between the machine machining electrode and the workpiece such that the inter-electrode mean machine machining voltage becomes a set voltage.

[0005] Patent literature 2 discloses a control device for a wire EDM machine and a machine learning device that can determine a correction parameter precisely and easily. The control device, which optimizes the correction parameter for the wire EDM process, comprises a machine learning device that learns the correction parameter for the wire EDM process. The machine learning device includes a state observation unit that monitors state data indicating the state of the wire EDM process as a state variable, a determination data acquisition unit that acquires determination data indicating the correction parameter for the case where machining precision in the wire EDM process is favorable, and a learning unit that learns the correction parameter in conjunction with the state of the wire EDM process based on the state variable and the determination data.

[0006] Patent reference 3 discloses a wire EDM machine in which wire electrodes are wound around guide rollers to position several parallel cutting wire sections relative to a workpiece. The wire EDM machine comprises a drive unit for adjusting the distance between the workpiece and the cutting wire sections, a processing power source that applies a pulsed AC voltage between the workpiece and the cutting wire sections, a processing state detection device that detects the processing state of the cutting wire sections, and a processing control unit that controls the drive unit and the processing power source. When a value indicating the processing state exceeds a threshold, the processing control unit issues a command to apply the pulsed AC voltage according to the processing conditions to prevent wire electrode breakage. List of patent literature Patent literature 1: JP 2020 - 146 788 A Patent literature 2: US 2018 / 0 281 091 A1 Patent literature 3: JP 6 991 414 B1 Brief description of the invention Problem to be solved by the invention

[0007] When estimating the inter-electrode distance using the voltage or frequency of electrical discharges, the voltage or frequency of the electrical discharges fluctuates at a high frequency. If feedback control is implemented by directly monitoring the voltage or frequency of the electrical discharges, the fluctuation of the control target becomes too large. Therefore, as in the conventional technique described above, a control system is used that is obtained by averaging these fluctuations. However, in such a control system, if a shape with a rapid change in the inter-electrode distance results from previous machining operations, shape correction may not be adequately performed due to a delay in the control system, and the shape variation may not be eliminated. This is problematic because the fluctuation in the relative motion speed is gradual.

[0008] The present disclosure was made in view of the above, and its purpose is to provide a control unit for a wire EDM machine which can perform shape correction machining with a higher accuracy than before, even when a sudden change in shape occurs as a result of the previous machining. Means to solve the problem

[0009] To solve the problems described above and to achieve the objective, the present disclosure is a control unit for a wire EDM machine which machine-machines a workpiece by applying a voltage between the workpiece and an electrode and causing an electrical discharge, wherein the control unit is configured to control a drive trajectory of the electrode relative to the workpiece, a relative machine-machining speed between the workpiece and the electrode and a frequency of electrical discharges of a voltage periodically applied between the electrode and the workpiece, wherein the control unit comprises a storage device and a computing device.The storage device stores at least one element of data selected from a voltage mean value, which is an average of a voltage applied between the workpiece and the electrode, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of an (n-1)th machine operation of n machine operations in a predetermined machine processing section of the workpiece, where n is an integer of two or more.The calculating device calculates an electrode gap, which is a distance between the workpiece and the electrode, of an nth machining operation using a calculation model that specifies a relationship between the data of the (n-1)th machining operation and a machining shape of the workpiece, and calculates at least one command value from the electrode gap of the nth machining operation, which is selected from the machining speed, the frequency of electrical discharges and the drive trajectory of the nth machining operation, which corresponds to a machining quantity required to achieve a desired shape, which is based on a machining program for machining the machining section. Effects of the invention

[0010] The control unit for the wire EDM machine according to the present disclosure can achieve the effect that a shape correction machine operation is carried out with a higher accuracy than before, even if a sudden change in shape occurs as a result of the previous machine operation. Brief description of the drawings Fig. Figure 1 is a block diagram showing an exemplary configuration of a wire EDM machine according to the first embodiment. Fig. Figure 2 is a block diagram showing an example of a PI control system. Fig. Figure 3 is a diagram showing an example of the relationship between the machining part shape and the machining speed in a conventional PI control system. Fig. Figure 4 is a diagram showing an example of the relationship between the machining part shape and the machining speed in the control system according to the first embodiment. Fig. Figure 5 is a diagram that schematically shows the amount of machining required on the workpiece at a corner section with a curved shape. Fig. Figure 6 is a diagram showing an example of the machining shape of the workpiece obtained as a result of the previous machining process. Fig. Figure 7 is a diagram showing an example of the machining shape of the workpiece obtained as a result of the previous machining. Fig. Figure 8 is a flowchart showing an exemplary procedure for the control method for the wire EDM machine according to the first embodiment. Fig. Figure 9 is a flowchart showing an exemplary procedure for the control method for the wire EDM machine according to the first embodiment. Fig. Figure 10 is a block diagram showing an exemplary configuration of a wire EDM machine according to the second embodiment. Fig. Figure 11 is a block diagram showing an exemplary configuration of a wire EDM machine according to the third embodiment. Fig. Figure 12 is a block diagram showing an exemplary configuration of a wire EDM machine according to the fifth embodiment. Fig. Figure 13 is a model diagram showing an example of the overview of machine learning according to the sixth embodiment. Fig. Figure 14 is a diagram which schematically shows an exemplary configuration of a learning device which is used in the control unit for the wire EDM machine according to the sixth embodiment. Fig. Figure 15 is a diagram that schematically shows an example of a neural network used by the model generation unit. Fig. Figure 16 is a flowchart showing an exemplary procedure for learning processing by the learning device according to the sixth embodiment. Fig. Figure 17 is a model diagram showing an example of the overview of machine learning according to the sixth embodiment. Fig. Figure 18 is a diagram which schematically shows an exemplary configuration of a sequencing device which is used in the control unit for the wire EDM machine according to the sixth embodiment. Fig. Figure 19 is a flowchart showing an exemplary procedure for inference processing by the inference device according to the sixth embodiment. Fig. Figure 20 is a diagram showing an exemplary configuration of the hardware of the control unit for the wire EDM machine according to the first to sixth embodiments. Description of embodiments

[0011] In the following, a control unit for a wire EDM machine and a control method for a wire EDM machine according to embodiments of the present disclosure are described in detail with reference to the drawings. First embodiment.

[0012] Fig. Figure 1 is a block diagram showing an exemplary configuration of a wire EDM machine according to the first embodiment. The wire EDM machine 1 comprises a machining electrode 10, a power supply unit 20, and a control unit 30. The power supply unit 20 and the control unit 30 constitute a control device for the wire EDM machine 1.

[0013] The wire EDM machine 1 is a machining device that repeatedly machine a predetermined section of the workpiece 11 n times. Here, n is an integer of two or more. The machining conditions, i.e., the distance between the workpiece 11 and the machining electrode 10 and the electrical energy, are changed with each machining operation. For example, a machining operation is performed such that the electrical energy decreases with an increasing number of machining operations. During the nth machining operation, a form correction operation is performed, which improves surface roughness and dimensional accuracy. Among the n machining operations, the first to (n-1)th machining operations are referred to as rough machining, and the nth machining operation is referred to as form correction machining.

[0014] The machining electrode 10 is an electrode formed from a wire-shaped conductive material, i.e., a wire electrode. Although the machining electrode 10 has a configuration that allows machining of the workpiece 11 with a wire, it exhibits Fig. Figure 1 shows a simplified configuration of the machining electrode 10. In this example, the machining electrode 10 is fed from the wire spool and its direction is changed by the feed roller, so that the machining electrode 10 is oriented vertically. The machining electrode 10 performs electrical discharge machining on the workpiece 11, passing through the hole of the upper guide head and the hole of the lower guide head. After passing through the lower guide head, the machining electrode 10 is changed in direction by the lower roller and collected by the collecting roller in the collection box.

[0015] The power supply unit 20 comprises a machine processing power supply 21 and a machine processing power supply control unit 22. The machine processing power supply 21 applies a voltage between the machine processing electrode 10 and the workpiece 11. The machine processing power supply control unit 22 controls the connection to and from the machine processing power supply 21. The wire EDM machine 1 applies a voltage between the workpiece 11 and the machine processing electrode 10 to generate an electrical discharge, thereby performing EDM machining on the workpiece 11. A detailed description of the power supply unit 20, including its mechanical structure, is omitted here because it is not the essential aspect of the present disclosure.

[0016] The control unit 30 controls the machine processing speed based on the average voltage, which is the average value between the machine processing electrode 10 and the workpiece 11 and is also referred to here as the inter-electrode average voltage. The machine processing speed is a relative speed between the machine processing electrode 10 and the workpiece 11.

[0017] The control unit 30 comprises a drive trajectory control unit 31, an inter-electrode mean voltage detection unit 32, a voltage calculation unit 33, a machine processing speed control unit 34, a drive control unit 35, an inter-electrode mean voltage storage unit 36, a frequency electrical discharge storage unit 37, a machine processing speed storage unit 38 and a drive trajectory storage unit 39.

[0018] The drive trajectory control unit 31 controls the movement of the axes of the wire EDM machine 1 according to the machining program. This means that the drive trajectory control unit 31 calculates the command value of the drive trajectory of the machining electrode 10 relative to the workpiece 11. For example, the drive trajectory control unit 31 determines from the machining program whether the machining profile of the machining section to be machined is straight or curved and calculates the drive trajectory, which is the path of the machining electrode 10. In the case of a curved profile, the drive trajectory is calculated using the corner diameter, which takes into account the radius of the corner section forming the curved profile, the diameter of the machining electrode 10, and the offset amount, as well as the opening angle of the corner section.The drive trajectory control unit 31 outputs the calculated command value of the drive trajectory to the drive control unit 35. The drive trajectory control unit 31 stores the drive trajectory in the drive trajectory storage unit 39.

[0019] The inter-electrode mean voltage detection unit 32 detects the inter-electrode mean voltage, that is, the average of the voltages between the machining electrode 10 and the workpiece 11 over a predetermined period. The inter-electrode mean voltage detection unit 32 stores the detected inter-electrode mean voltage in the inter-electrode mean voltage storage unit 36. At each predetermined period, the inter-electrode mean voltage detection unit 32 also estimates the state of electrical discharges, including the frequency of electrical discharges and the machining quantity, from the inter-electrode mean voltage and stores the estimated frequency of electrical discharges in the frequency electrical discharge storage unit 37.

[0020] The voltage calculation unit 33 calculates the difference between the detected mean inter-electrode voltage and a set voltage. During machining with the wire EDM machine 1, the suitable state of electrical discharges, including the frequency of electrical discharges and the amount of machining, varies depending on the machining conditions, and a suitable mean inter-electrode voltage is determined according to the purpose. Therefore, the set voltage is a voltage considered suitable and is pre-set according to the purpose.

[0021] The machine processing speed control unit 34 comprises a machine processing speed calculation unit 341, which calculates the machine processing speed, and a calculation device 342, which calculates a correction value to correct the machine processing speed calculated by the machine processing speed calculation unit 341. The machine processing speed calculation unit 341 calculates the machine processing speed such that the mean inter-electrode voltage measured within the predetermined time period becomes the set voltage, that is, the difference calculated by the voltage calculation unit 33 becomes zero. A known method can be used for calculating the machine processing speed by the machine processing speed calculation unit 341.According to an example, the machine processing speed calculation unit 341 can calculate the command value of the machine processing speed using at least one element of data from the inter-electrode mean voltage, the frequency of electrical discharges, and the drive trajectory of the current machine operation, which is the nth machine operation. It should be noted that the machine processing speed calculation unit 341 corresponds to a command value calculation unit.

[0022] If the machining section to be machined has a straight shape, the machining speed calculation unit 341 calculates the machining speed using proportional-integral (PI) control such that the difference between the inter-electrode mean voltage and the set voltage becomes zero. If the machining section has a curved shape, the machining speed calculation unit 341 calculates the machining speed taking into account the corner diameter of the corner section forming the curved shape at the drive position and the opening angle of the corner section, in addition to the inter-electrode mean voltage and the set voltage.

[0023] In a form correction machine operation that improves surface roughness and form accuracy, the machine processing speed calculation unit 341 further corrects the calculated machine processing speed using the correction value calculated by the calculation device 342. Hereinafter, the machine processing speed corrected by the correction value is referred to as the post-correction machine processing speed when it is distinguishable from the uncorrected machine processing speed. Furthermore, the uncorrected machine processing speed and the corrected machine processing speed are referred to as the machine processing speed when they are not distinguished from one another.

[0024] In a shape correction machine operation that improves surface roughness and shape accuracy, the calculation device 342 calculates the inter-electrode distance of the current machine operation using a calculation model that specifies the relationship between at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine operation speed and the drive trajectory of the previous machine operation, and the machine operation shape of the workpiece 11.Furthermore, the calculating device 342 calculates the command value of the machine processing speed for the current machine operation from the inter-electrode distance of the current machine operation. This command value corresponds to the machine processing quantity required to achieve the desired shape, which is based on the machine operation program for machining the machine operation section. If the machine operation is performed n times in the machine operation section, the previous machine operation corresponds to the (n-1)th machine operation, and the current machine operation corresponds to the nth machine operation. At least one element of data, selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the previous machine operation, is, for example, data that specifies a profile.The calculation model indicates the relationship between the above data and the machining mode of workpiece 11, but can also indicate the relationship between the above data and the inter-electrode distance of the current machining operation.

[0025] The machine processing speed calculated by the machine processing speed calculation unit 341 serves to bring the calculated machining shape of workpiece 11, which is obtained by performing the (n-1)th machining operation based on the machine processing program, into the desired shape. The calculated machining shape of workpiece 11 often does not correspond to the actual machining shape of workpiece 11.The command value for machine processing speed calculated by the calculating device 342 is a command value for eliminating the difference between the actual machining shape of workpiece 11, obtained as a result of the previous machining operation, and the desired machining shape. It is also a correction value for correcting the machining speed calculated by the machining speed calculation unit 341. The command value for machine processing speed calculated by the calculating device 342 is therefore also referred to as a correction value. Technically, the actual machining shape of workpiece 11, obtained as a result of the previous machining operation, is also a machining shape that is estimated by calculation.Because the estimation is performed using at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the previous machine operation, the machine operation shape estimated using the above data is closer to the actual machine operation shape than the calculated machine operation shape of workpiece 11 obtained by performing the (n-1)th machine operation based on the machine operation program.

[0026] To estimate the inter-electrode distance, a computational model is obtained beforehand, which describes the relationship between at least one element of data selected from the mean inter-electrode voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation, and the machining shape of the workpiece 11. Using a function that incorporates this computational model, the inter-electrode distance of the current machining operation is then calculated, and furthermore, the command value of the machine processing speed to achieve the desired shape is calculated.Regarding the calculation model, the accuracy of the estimated inter-electrode distance increases with an increasing number of data elements selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation. It is therefore desirable to use as many types of data as possible from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation. The machine processing speed control unit 34 outputs the calculated command value of the machine processing speed to the drive control unit 35.

[0027] The drive control unit 35 controls the operation of the machining electrode 10 relative to the workpiece 11 based on the drive trajectory command value from the drive trajectory control unit 31 and the machining speed command value from the machining speed control unit 34. This means that the position of the machining electrode 10 relative to the workpiece 11 is controlled based on the drive trajectory command value, and that the relative speed of the machining electrode 10 relative to the workpiece 11 is controlled based on the machining speed command value. The drive control unit 35 can move the workpiece 11 based on the command values, move the machining electrode 10 based on the command values, or move both the workpiece 11 and the machining electrode 10 based on the command values.

[0028] The inter-electrode mean voltage storage unit 36 ​​stores the value of the inter-electrode mean voltage, which is detected by the inter-electrode mean voltage detection unit 32 at each predetermined time interval. As described above, the wire EDM machine 1 continuously machines the machining section n times. Then, on the last, nth machining operation, shape correction machining is performed, which increases the accuracy of the shape according to the desired form and reduces the surface roughness. This means that a machining operation is performed repeatedly multiple times under varying machining conditions.In the first embodiment, the inter-electrode average voltage storage unit 36 ​​only needs to be able to store at least the inter-electrode average voltage of the machining operation immediately preceding the form correction machining operation, i.e., the (n-1)th machining operation. According to one example, the inter-electrode average voltage storage unit 36 ​​stores the value of the inter-electrode average voltage in association with the coordinate value of the drive trajectory at the time of each machining operation. Furthermore, the inter-electrode average voltage storage unit 36 ​​can store a load-free time, which is the duration from the application of a voltage until the occurrence of an electrical discharge; however, in this description, it stores the inter-electrode average voltage. The inter-electrode average voltage can be measured by the machining power supply control unit 22.

[0029] The frequency-electrical-discharge storage unit 37 stores the value of the frequency of electrical discharges, which is estimated by the inter-electrode mean voltage detection unit 32. The frequency-electrical-discharge storage unit 37 only needs to be able to store at least the frequency of electrical discharges from the machining operation immediately preceding the form correction machine operation, i.e., the (n-1)th machining operation. For example, the frequency-electrical-discharge storage unit 37 stores the frequency of electrical discharges in association with the coordinate value of the drive trajectory at the time of each machining operation. The frequency of electrical discharges can be measured by the machining power supply control unit 22.

[0030] The machine processing speed storage unit 38 stores the command value of the machine processing speed, which is calculated by the machine processing speed control unit 34. The machine processing speed storage unit 38 must only be able to store at least the machine processing speed of the machining operation immediately preceding the form correction machine operation, i.e., the (n-1)th machining operation. For example, the machine processing speed storage unit 38 stores the command value of the machine processing speed in association with the coordinate value of the drive trajectory at the time of each machining operation.

[0031] The drive trajectory storage unit 39 stores the drive trajectory from the drive trajectory control unit 31. If the machining section has a straight shape, a coordinate value relative to a predetermined position is stored as the drive trajectory. If the machining section is a corner section, in addition to the coordinate value, the drive trajectory at the corner section, including the radius and the opening angle of the corner section, is also stored. The drive trajectory storage unit 39 only needs to be able to store at least the drive trajectory of the machining operation immediately preceding the shape correction machining operation, i.e., the (n-1)th machining operation.

[0032] The Inter-Electrode Mean Voltage Storage Unit 36, the Frequency Electric Discharge Storage Unit 37, the Machine Processing Speed ​​Storage Unit 38 and the Drive Trajectory Storage Unit 39 correspond to one storage device.

[0033] The control of the machine processing speed for a straight shape and the control of the machine processing speed for a curved shape in shape correction machine machining are now described.

[0034] First, the control of the machine processing speed for a straight shape is described. In the wire EDM machine 1, the power supply unit 20 applies a voltage between the electrodes to continuously generate an electrical discharge for machining. The inter-electrode mean voltage detection unit 32 measures the inter-electrode mean voltage at each predetermined time interval. Furthermore, the inter-electrode mean voltage detection unit 32 can estimate the state of electrical discharges, including the frequency of electrical discharges and the amount of machining performed, from the inter-electrode mean voltage at each predetermined time interval.The frequency of electrical discharges, the amount of machining, and similar factors considered suitable vary depending on the machining conditions, and a suitable inter-electrode mean voltage is determined according to the purpose. This means that, for the set voltage, a voltage considered suitable is preset according to the purpose. The voltage calculation unit 33 takes the inter-electrode mean voltage measured by the inter-electrode mean voltage detection unit 32 and calculates the difference between the inter-electrode mean voltage and the set voltage. The voltage calculation unit 33 outputs the calculated difference to the machine machining speed control unit 34.

[0035] When the difference is taken from the voltage calculation unit 33, the machine processing speed calculation unit 341 of the machine processing speed control unit 34 calculates the machine processing speed so that the difference calculated by the voltage calculation unit 33 becomes zero. The drive trajectory control unit 31 calculates the command value of the drive trajectory according to the machine processing program. The drive control unit 35 controls the operation of the machine processing electrode 10 based on the command value of the machine processing speed calculated by the machine processing speed control unit 34 and the command value of the drive trajectory calculated by the drive trajectory control unit 31.

[0036] If the machine processing speed calculation unit 341 of the machine processing speed control unit 34 directly converts the voltage calculated from the difference between the inter-electrode mean voltage with a large fluctuation and the set voltage into a speed, rapid speed fluctuations occur, and form accuracy and surface roughness deteriorate. Therefore, the machine processing speed calculation unit 341 obtains the machine processing speed using a PI control system, such as the one described in Fig. 2 shown. Fig. Figure 2 is a block diagram showing an example of a PI control system. As shown in Fig. As shown in Figure 2, in the PI control system proportional control is carried out by multiplying the difference between the inter-electrode mean voltage and the set voltage by a predetermined proportional gain, integral control is carried out by multiplying the integral gain according to the accumulated amount of deviation, and the machine processing speed is determined by adding these results.

[0037] Fig. Figure 3 is a diagram showing an example of the relationship between the machining part shape and the machining speed in a conventional PI control system. Fig. Figure 3 shows a case in which a small irregularity exists in the machined part shape, which is the resulting shape of the previous machining operation. In a PI control system, such as the one in Fig. As shown in Figure 2, if the gain of the integral term is small, the integral term as a whole reacts to the presence of a frequency component, which weakens the integral term and slows down the reaction. For this reason, in the case of machining the workpiece 11 with the method shown in Figure 2, the integral term is treated as follows: Fig. In the example of the machined part shape shown, time is required until the machine processing speed, calculated from the mean inter-electrode voltage, reaches a speed at which the unevenness can be corrected, and the shape cannot be sufficiently corrected. This means that the machined part shape cannot be sufficiently corrected due to a delay in the speed control system. In the example of Fig. 3. The machine processing speed decreases even after the apex of the machined part's shape has been reached. Furthermore, the machine processing speed increases sharply after the apex of the machined part's shape, leading to oscillation. In this way, the shape correction machine, which is based solely on PI control, cannot produce the desired shape.

[0038] Therefore, in the control system according to the first embodiment, the position of the previous machine operation, which is carried out before the form correction machine operation, is stored in the drive trajectory storage unit 39, the inter-electrode mean voltage is stored in the inter-electrode mean voltage storage unit 36, the frequency of electrical discharges is stored in the frequency of electrical discharges storage unit 37, and the machine operation speed is stored in the machine operation speed storage unit 38.Then the calculation device 342 of the machine processing speed control unit 34 calculates the inter-electrode distance of the current machine processing of the position in advance from the profiles of the inter-electrode mean voltage, the frequency of electrical discharges and the machine processing speed of the previous machine processing of the position using the calculation model and calculates the correction value of the machine processing speed to achieve the desired machine processing shape from the pre-calculated inter-electrode distance.As described above, the calculation model includes a function for calculating the inter-electrode distance between the machining shape of the workpiece 11 of the previous machining operation and the machining electrode 10 of the current machining operation from the profiles of the inter-electrode mean voltage, the frequency of electrical discharges and the machining speed of the previous machining operation.Furthermore, the calculation model includes a function for specifying the quantity of workpiece 11 to be removed in the current machine operation using the pre-calculated inter-electrode distance and the distance between the desired machining shape, which is obtained based on the machine operation program, and the machine operation electrode 10 of the current machine operation, and for calculating the machine operation speed of the machine operation electrode 10 to remove this quantity of workpiece 11.

[0039] Fig. Figure 4 is a diagram illustrating an example of the relationship between the machining part shape and the machining speed in the control system according to the first embodiment. In the first embodiment, the inter-electrode distance is calculated, but the diagram shows... Fig. 4. The machining part shape obtained when the inter-electrode distance is calculated is described to facilitate understanding of the description. As in the machining part shape in Fig. As shown in Figure 4, in the first embodiment, a protruding shape in the machining section is pre-calculated. Therefore, the computational device 342 calculates the correction value of the machining speed such that the command value of the machining speed decreases before the protruding section and reduces the post-correction machining speed calculated by the machining speed computational unit 341. Furthermore, the computational device 342 calculates the correction value of the machining speed such that the machining speed does not oscillate after passing through the section with the protruding section. This makes it possible to correct the unevenness on the machined surface of the workpiece 11 without reducing the integral gain or the proportional gain of the PI control system. Fig. to increase by 2. In addition, it takes place in Fig. 4. The machining speed decreases even after passing the apex of the protruding section. This is due to the fact that machining is performed not only by an electrical discharge at the front of the machining electrode 10, but also by an electrical discharge at the rear. This means that the amount of machining at the front and rear of the machining electrode 10, which has passed the apex of the protrusion, is greater than when the machining electrode 10 is at the apex of the protruding section, and therefore the machining speed is reduced. Points to consider in Fig. Four are that the machine processing speed is faster than in the case of Fig. 3 decreases and that the speed does not increase significantly, i.e., it does not exhibit damped oscillation after the machined part shape returns to a straight line after the protruding section.

[0040] Given the proportional gain as K1, the integral gain as K2, and the set voltage as V s , the inter-electrode mean voltage as V ave and the correction value of the machine processing speed, which is calculated based on the machine processing part shape of the previous machine processing, as F comp The formula of the PI control system for calculating the post-correction machine processing speed F is expressed by formula (1) below. The correction value F compThe machine processing speed is calculated using the profiles of the previous machine processing, which are stored in the inter-electrode mean voltage storage unit 36, the frequency electrical discharge storage unit 37, and the machine processing speed storage unit 38. The inter-electrode mean voltage of the previous machine processing, which is stored in the inter-electrode mean voltage storage unit 36, is represented as V n-1 , the frequency of electrical discharges from the previous machine operation, which is stored in the frequency-electrical-discharge storage unit 37, as S n-1 and the machine processing speed of the previous machine processing, which is stored in the machine processing speed memory 38, as F n-1 , given, then the correction value F compthe machine processing speed is expressed in particular by formula (2) below. F=K1(Vave−Vs)+∫K2(Vave−Vs)dt+FcompFcomp=∫f(Vn−1,Fn−1,Sn−1)dt

[0041] As described above, formula (2) is a function which provides a calculation model for obtaining the profiles of the inter-electrode mean voltage V. n-1 , the frequency of electrical discharges S n-1 and the machine processing speed F n-1 the previous machining process and to calculate the inter-electrode distance of the current machining process from the profiles of the inter-electrode mean voltage V n-1 , the frequency of electrical discharges S n-1 and the machine processing speed F n-1 It includes, and calculates the machine processing speed from the inter-electrode distance.

[0042] Next, the control of the machine processing speed for a curved shape will be described. Fig. Figure 5 is a diagram schematically illustrating the machining volume of the workpiece at a corner section with a curved shape. The case shown is the machining of the inner corner, i.e., the inside of the corner section. It is known that the machining volume at the inner corner, with the machining electrode 10 rotating θ from the center of curvature of the corner section at the circumferential position of the corner diameter R, is expressed by formula (3) below. Here, R is the corner diameter, taking into account the diameter and offset of the machining electrode 10, and θ is the opening angle of the corner section. The distance between the center of the machining electrode 10 and the surface of the workpiece 11 after machining is represented by GAP. Machine processing quantity at inner corner = R*LU¨CKE*θ + LU¨CKE2*θ

[0043] In equation (3), the machining electrode 10 is moved by R*θ. The machining quantity with the movement R*θ during the machining of a straight shape is expressed by equation (4) below. Machine processing quantity for straight shape = R*LUCKE*θ

[0044] Formulas (3) and (4) indicate that the machining quantity at the inner corner is greater than the machining quantity at the straight shape. This means for the corner shape that, as in Fig. As shown in Figure 5, the required machine processing quantity, i.e., the area facing the machine processing electrode 10, changes compared to the straight shape, and it is therefore necessary to change the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the like as a function of the change in the machine processing quantity.

[0045] Patent literature 1 discloses a method for calculating the machine processing speed and the frequency of electrical discharges according to a change in the amount of machine processing applied to a corner section. However, if such forward control is performed as a function of the change in the amount of machine processing, and if the result of the previous machine processing is not an exact corner shape, the desired corner shape cannot be obtained with the current shape correction machine processing.

[0046] In the control system according to the first embodiment, as in the case of the straight form, the position of the previous machining operation is stored in the drive trajectory storage unit 39, the inter-electrode mean voltage is stored in the inter-electrode mean voltage storage unit 36, the frequency of electrical discharges is stored in the frequency-electrical discharge storage unit 37, and the machining speed is stored in the machining speed storage unit 38. Furthermore, for machining the corner section, information about the corner section is also stored in the drive trajectory storage unit 39, along with the position of the previous machining operation.The information about the corner section includes the drive trajectory at the corner section, i.e., the radius and opening angle of the corner section, and the like. The calculation device 342 of the machine processing speed control unit 34 calculates a correction value for adjusting the machine processing speed at the position of the corner section of the current machine operation from the information about the corner section, which is stored in the drive trajectory storage unit 39, in addition to the profiles of the frequency of electrical discharges, the inter-electrode mean voltage, and the machine processing speed of the previous machine operation.

[0047] Fig. 6 and Fig. Figure 7 shows diagrams illustrating examples of the workpiece's machining shape as a result of previous machining operations. In these drawings, dashed lines indicate the desired shape, and dashed lines indicate the shape actually obtained through previous machining. Fig. Figure 6 shows a case in which the actual corner diameter obtained is smaller than the shape of the desired corner section. Fig. Figure 7 shows a case in which there is an unevenness in the entrance / exit of the corner section. In the first embodiment, the calculating device 342 calculates the inter-electrode distance of the current machine operation, which corresponds to the currently obtained shape in Fig. 6 and Fig. 7, using the calculation model, calculates the machine processing speed required to remove the area representing the difference between the actual shape obtained and the desired shape. Fig. 6 and Fig. 7 corresponds to using the inter-electrode distance and sets the machine processing speed as the correction value. By performing machining at the post-correction machine processing speed, which has been corrected with the correction value, it is possible to perform machining that achieves the desired shape. As in Fig. 6 and Fig. As shown in Figure 7, if the shape of the corner section is smaller than the desired shape as a result of previous machining, or if there is an unevenness in the entrance / exit of the corner section, it is possible to improve the shape accuracy after machining by optimizing the machining speed depending on these shapes.

[0048] A formula for calculating the post-correction machine processing speed F cnr The operation at the corner section is expressed by formula (5) below. It should be noted that the position at which machining is performed on the corner section is expressed by x. Furthermore, the function g is a machining speed that takes into account a change in the amount of machining performed on the corner section, compared to the case of the straight shape, and can be obtained using a known method. The correction value F comp_cnrThe machine processing speed is calculated using formula (6) below, using the corner diameter, opening angle and position on the corner section in addition to the profiles of the previous machine processing, which are stored in the Inter-Electrode Mean Voltage Storage Unit 36, the Frequency Electrical Discharge Storage Unit 37 and the Machine Processing Speed ​​Storage Unit 38. Fcnr=g(R,θ,Vs,Vave,x)+Fcomp_cnr Fcomp_cnr=∫h(R,θ,Vn−1,Fn−1,Sn−1,x)dt

[0049] As shown in equations (5) and (6), the required machining quantity at the corner section changes compared to the straight section depending on the position of the machining electrode 10 at the corner section, and therefore the position at the corner section is included in the function. Equation (5) describes a case in which the function g corrects the difference in machining quantity between the corner section and the straight section by controlling the machining speed. Alternatively, the function g can correct the difference in machining quantity between the corner section and the straight section by controlling the frequency of electrical discharges or the trajectory.Although the above description shows an example of machining the inner corner, the machining speed can also be controlled in a similar way when machining the outer corner outside the corner section. In the case of machining the outer corner, the machining volume is smaller than in the case of straight machining.

[0050] Next, a control procedure for this wire EDM machine 1 will be described. Fig. 8 and Fig. Figure 9 shows a flowchart illustrating an exemplary procedure for the control method of the wire EDM machine according to the first embodiment. First, the drive trajectory control unit 31 calculates the drive trajectory from the machine program and stores it in the drive trajectory storage unit 39 (step S11). The drive trajectory control unit 31 outputs the calculated drive trajectory to the drive control unit 35. The drive control unit 35 drives the machine electrode 10 based on the command value of the drive trajectory (step S12). Then, the machine power supply control unit 22 controls the machine power supply 21 to apply a voltage between the electrodes with a predetermined frequency of electrical discharges (step S13).The inter-electrode mean voltage detection unit 32 then measures the inter-electrode mean voltage at each predetermined time interval and stores the inter-electrode mean voltage, which is the measurement result, in the inter-electrode mean voltage storage unit 36 ​​(step S14). The inter-electrode mean voltage detection unit 32 estimates the frequency of electrical discharges from the inter-electrode mean voltage at each predetermined time interval and stores the estimated frequency of electrical discharges in the frequency electrical discharge storage unit 37 (step S15).

[0051] Next, it is determined whether straight machining is being performed (step S16). The drive trajectory control unit 31 determines whether straight or curved machining is being performed when determining the drive trajectory. In response to a determination that straight machining is being performed (yes in step S16), the voltage calculation unit 33 calculates the difference between the inter-electrode mean voltage, which is obtained from the inter-electrode mean voltage detection unit 32, and the set voltage (step S17). The voltage calculation unit 33 outputs the calculated difference to the machine speed control unit 34.The machine processing speed calculation unit 341 of the machine processing speed control unit 34 calculates the machine processing speed at which the difference becomes zero through IP control (step S18).

[0052] In response to a determination in step S16 that no straight machining operation is performed, that is, in response to a determination that curved machining operation is performed (no in step S16), the machining speed calculation unit 341, on the other hand, calculates the machining speed using the corner diameter R, the opening angle θ of the corner section, and the set speed V. S, the inter-electrode mean voltage of the current machine operation and the position x at the corner section (step S19). According to an example, the machine operation speed calculation unit 341 calculates the machine operation speed at which the difference between the inter-electrode mean voltage and the set voltage becomes zero, by taking into account the corner diameter R and the opening angle θ of the corner section at the position x of the current machine operation.

[0053] Afterwards, or after step S18, the machine processing speed calculation unit 341 determines whether a shape correction machining operation is performed (step S20). In response to a determination that a shape correction machining operation is not performed (no in step S20), the machine processing speed calculation unit 341 stores the calculated machine processing speed in the machine processing speed storage unit 38 (step S21). The drive trajectory control unit 31 calculates the drive trajectory and stores the calculated drive trajectory in the drive trajectory storage unit 39 (step S22). Next, the drive control unit 35 drives the machine processing electrode 10 based on the drive trajectory and the command value of the machine processing speed (step S23).The process then returns to step S13 and a rough machining operation, which is not a shape correction machine operation, is carried out.

[0054] In response to a determination in step S20 that a shape correction machining operation is performed (yes in step S20), the computation device 342 of the machine machining speed control unit 34, on the other hand, uses formula (2) or formula (6) to calculate the inter-electrode distance of the current machining operation from at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine machining voltage, and the drive trajectory during the previous machining operation, and calculates from the inter-electrode distance the correction value of the machine machining speed to bring the workpiece 11 into the desired machining shape (step S24).Next, the machine processing speed calculation unit 341 uses formula (1) or formula (5) to calculate the post-correction machine processing speed, which is obtained by correcting the calculated machine processing speed with the correction value (step S25). Then, the drive trajectory control unit 31 calculates the drive trajectory (step S26). Finally, the drive control unit 35 performs a form correction machining operation based on the drive trajectory and the post-correction machine processing speed (step S27), and the process ends.

[0055] In the first embodiment, at least one element of data is stored, selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the (n-1)th machining operation of the n machining operations in the machining section, where n is an integer of two or more. The inter-electrode distance of the nth machining operation is calculated using a computational model that specifies the relationship between the data of the (n-1)th machining operation and the machining shape of the workpiece 11. The correction value of the machine processing speed of the nth machining operation, which corresponds to the machining quantity required to achieve the desired shape, is calculated from the inter-electrode distance of the nth machining operation.This means that the machining shape of workpiece 11 after the (n-1)th machining operation, which is the previous machining operation, can be estimated with high accuracy. Therefore, the machining operation can be carried out in such a way that the workpiece 11 has the desired shape during the shape correction machining operation. As a result, it is possible to perform shape correction machining with high accuracy even if a sudden change in shape occurs as a result of the previous machining operation.

[0056] In particular, at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the (n-1)th machine operation is data that reflects the state of the (n-1)th machine operation, and, as described above, the machining shape of the workpiece 11 of the (n-1)th machine operation can be estimated using this data. This means that in the first embodiment, in order to accurately estimate the machining shape of the workpiece 11 of the (n-1)th machine operation, at least one element of data is stored that is selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the (n-1)th machine operation. Second embodiment.

[0057] Fig. Figure 10 is a block diagram showing an exemplary configuration of a wire EDM machine according to the second embodiment. It should be noted that components identical to those of the first embodiment are designated by the same reference numerals, their descriptions omitted, and differences compared to the first embodiment are described.

[0058] The wire EDM machine 1a according to the second embodiment comprises a machine processing speed control unit 34a instead of the machine processing speed control unit 34 and a machine processing power supply control unit 22a instead of the machine processing power supply control unit 22.

[0059] The machine processing speed control unit 34a corresponds to the machine processing speed calculation unit 341, which is described in the first embodiment, and does not include the calculation device 342. This means that the machine processing speed control unit 34a calculates the machine processing speed in such a way that the difference calculated by the voltage calculation unit 33 becomes zero.

[0060] The machine processing power supply control unit 22a comprises a frequency-electrical discharge control unit 221, which controls the machine processing power supply 21 at a predetermined frequency of electrical discharges according to the machine processing conditions, and a calculation unit 222, which calculates a correction value to correct the predetermined frequency of electrical discharges according to the machine processing conditions. The frequency-electrical discharge control unit 221 controls the on / off cycle of the machine processing power supply 21. It should be noted that the frequency-electrical discharge control unit 221 can calculate the command value of the frequency of electrical discharges using at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, and the drive trajectory of the current machine processing.In a form correction machine operation that improves surface roughness and form accuracy, the frequency-electrical discharge control unit 221 corrects the predetermined frequency of electrical discharges with the correction value calculated by the calculation unit 222 and controls the machine processing power supply 21 at the post-correction frequency of electrical discharges. The frequency-electrical discharge control unit 221 corresponds to a command value calculation unit.

[0061] In shape correction machine machining, which improves surface roughness and shape accuracy, the calculation device 222 calculates the inter-electrode distance of the current machine machining using a calculation model that specifies the relationship between at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine machining speed and the drive trajectory of the previous machine machining, and the machine machining shape of the workpiece 11.Furthermore, the computing device 222 calculates the command value of the electrical discharge frequency of the current machining operation from the inter-electrode distance of the current machining operation. This frequency corresponds to the machining quantity required to achieve the desired shape, which is based on the machining program for the machining section. If a machining operation is performed n times in the machining section, the previous machining operation corresponds to the (n-1)th machining operation, and the current machining operation corresponds to the nth machining operation. At least one element of the data, selected from the inter-electrode mean voltage, the electrical discharge frequency, the machining speed, and the drive trajectory of the previous machining operation, is, for example, data that specifies a profile.The frequency of electrical discharges is controlled using the electrical discharge pause time of the machine processing power supply control unit 22a. It is assumed here that the computing device 222 calculates the command value of the electrical discharge pause time as the frequency of electrical discharges.

[0062] The predetermined frequency of electrical discharge serves to bring the calculated machining shape of workpiece 11, obtained by performing the (n-1)th machining operation based on the machining program, into the desired shape. The calculated machining shape of workpiece 11 often does not correspond to the actual machining shape of workpiece 11. The command value of the electrical discharge frequency, which is calculated by the computing device 222, is a command value for eliminating the difference between the actual machining shape of workpiece 11, obtained as a result of the previous machining operation, and the desired machining shape, and is also a correction value for correcting the predetermined frequency of electrical discharges.Therefore, the command value for the frequency of electrical discharges calculated by the calculating device 222 is also referred to as a correction value. The calculating device 222 outputs the correction value to the frequency-electrical-discharge control unit 221.

[0063] To estimate the inter-electrode distance, a computational model is obtained beforehand, which describes the relationship between at least one element of data selected from the mean inter-electrode voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation, and the machining shape of the workpiece 11. Using a function that incorporates this computational model, the command value of the electrical discharge frequency required to achieve the desired shape is then calculated.Regarding the calculation model, the accuracy of the estimated inter-electrode distance increases with an increasing number of data elements selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machine operation. It is therefore preferable to use as many types of data as possible from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machine operation. The machine processing power supply control unit 22a controls the machine processing power supply 21 using the calculated command value of the electrical discharge pause time.

[0064] In particular, the calculating device 222 uses at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation to estimate the inter-electrode distance of the current machining operation at the position that reflects the shape of the workpiece 11 from the previous machining operation. The calculating device 222 calculates the correction value for the electrical discharge pause time in addition to the estimated inter-electrode distance result so that the desired shape is obtained from the inter-electrode distance. For example, if the corresponding position is pre-existing as a result of the previous machining operation, that is, if the inter-electrode distance is small, the electrical discharge pause time is calculated to be short in order to increase the machining volume.Furthermore, if the corresponding position is deepened as a result of previous machining, that is, if the inter-electrode distance is large, the electrical discharge pause time is calculated as long in order to reduce the amount of machining or not to perform the machining.

[0065] When controlling the machine processing speed for a straight form, a formula is used to calculate the post-correction electrical discharge pause time (OFF). comp as expressed by formula (7) below. Here, OFF is the set electrical discharge pause quantity during machine operation, and C1 is the conversion coefficient for the correction value of the machine operation speed into the correction value of the electrical discharge pause quantity. The frequency electrical discharge control unit 221 calculates the command value of the post-correction electrical discharge pause time OFF. compBy adding the pause quantity, which is the correction value calculated by the calculation device 222, to the pause time OFF, which is determined in advance according to machine processing conditions. By pre-calculating the shape and controlling the frequency of electrical discharges in this way, it is possible to correct the unevenness on the machined surface without reducing the integral gain or increasing the proportional gain of the PI control system. OFFcomp=OFF+C1*∫f(Vn−1,Fn−1,Sn−1)dt

[0066] Even when controlling the machine processing speed for a curved shape, the difference between the desired shape and the machine processing shape of the corner section, estimated as a result of previous machining, can be eliminated by controlling the frequency of electrical discharges. Specifically, the computation unit 222 calculates the command value of the electrical discharge frequency at the corner section position from the corner section information stored in the drive trajectory memory unit 39, in addition to the profiles of previous machining stored in the interelectrode mean voltage memory unit 36, the electrical discharge frequency memory unit 37, and the machine processing speed memory unit 38. As described above, the corner section information includes the corner diameter, the opening angle of the corner section, and the like.A formula for calculating the post-correction electrical discharge pause quantity OFF. comp_cnr The corner section is expressed by formula (8) below. Here, the conversion coefficient for the correction value of the machine processing speed into the correction value of the electrical discharge interval is represented by C2. OFFcomp_cnr=OFF+C2*∫h(R,θ,Vn−1,Fn−1,Sn−1,x)dt

[0067] As in Fig. 6 and Fig. As shown in Figure 7, it is possible to improve the shape accuracy after machining by calculating the command value of the frequency of electrical discharges according to these shapes if the shape of the corner section as a result of the previous machining is smaller than the desired shape or if there is an unevenness in the entrance / exit of the corner section.

[0068] In the second embodiment, at least one element of data is stored, selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the (n-1)th machining operation of n machining operations in the machining section, where n is an integer of two or more. The inter-electrode distance of the nth machining operation is calculated using a computational model that specifies the relationship between the data of the (n-1)th machining operation and the machining shape of the workpiece 11. The command value of the frequency of electrical discharges of the nth machining operation, which corresponds to the machining quantity required to achieve the desired shape, is calculated from the inter-electrode distance of the nth machining operation.This means that the machining shape of workpiece 11 after the (n-1)th machining operation, which is the previous machining operation, can be estimated with high accuracy, and therefore the machining operation can be carried out in such a way that the workpiece 11 has the desired shape in the shape correction machining operation. Accordingly, it is possible to carry out shape correction machining with high accuracy even if a sudden change in shape occurs as a result of the previous machining operation. Third embodiment.

[0069] Fig. Figure 11 is a block diagram showing an exemplary configuration of a wire EDM machine according to the third embodiment. It should be noted that components identical to those of the first embodiment are designated by the same reference numerals, their descriptions omitted, and differences compared to the first embodiment are described.

[0070] The wire EDM machine 1b according to the third embodiment comprises a machine processing speed control unit 34b instead of the machine processing speed control unit 34 and a drive trajectory control unit 31b instead of the drive trajectory control unit 31.

[0071] The machine processing speed control unit 34b corresponds to the machine processing speed calculation unit 341, which is described in the first embodiment, and does not include the calculation device 342. This means that the machine processing speed control unit 34b calculates the machine processing speed in such a way that the difference calculated by the voltage calculation unit 33 becomes zero.

[0072] The drive trajectory control unit 31b comprises a drive trajectory calculation unit 311, which calculates a drive trajectory, and a calculation device 312, which calculates a correction value for correcting the drive trajectory calculated by the drive trajectory calculation unit 311. The drive trajectory calculation unit 311 calculates the command value of the drive trajectory of the machining electrode 10 according to the machining program. Furthermore, the drive trajectory calculation unit 311 can calculate the command value of the drive trajectory using at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, and the drive trajectory of the current machining operation.In a form correction machine operation that improves surface roughness and form accuracy, the drive trajectory calculation unit 311 corrects the calculated drive trajectory with the correction value calculated by the calculation device 312 and controls the machine machining electrode 10 with the post-correction drive trajectory. The drive trajectory calculation unit 311 corresponds to a command value calculation unit.

[0073] In a shape correction machine operation that improves surface roughness and shape accuracy, the calculation device 312 calculates the inter-electrode distance of the current machine operation using a calculation model that specifies the relationship between at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine operation speed and the drive trajectory of the previous machine operation, and the machine operation shape of the workpiece 11.Furthermore, the calculating device 312 calculates the command value of the drive trajectory of the current machining operation from the inter-electrode distance of the current machining operation. This command value corresponds to the machining quantity required to achieve the desired shape, which is based on the machining program for machining the machining section. The drive trajectory calculated here is an axis movement trajectory, which is a trajectory along which the axis of the wire EDM machine 1b moves. If the machining operation is performed n times in the machining section, the previous machining operation corresponds to the (n-1)th machining operation, and the current machining operation corresponds to the nth machining operation.At least one element of the data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the previous machine operation is, for example, data that specifies a profile. Hereinafter, the command value of the drive trajectory of the current machine operation, calculated by the computation unit 312, is also referred to as an axis movement trajectory. The axis movement trajectory is calculated in a normal vector direction that is perpendicular to the machine processing direction vector, which is obtained from the drive trajectory calculated by the drive trajectory computation unit 311. It should be noted that the normal vector direction is a direction perpendicular to the machine processing direction vector and is not a direction of extension of the machine processing electrode 10.

[0074] The drive trajectory calculated by the drive trajectory calculation unit 311 serves to transform the calculated machining shape of workpiece 11, obtained by performing the (n-1)th machining operation based on the machining program, into the desired shape. The calculated machining shape of workpiece 11 often does not correspond to the actual machining shape of workpiece 11. The command value of the axis motion trajectory, calculated by the calculation device 312, is a command value for eliminating the difference between the actual machining shape of workpiece 11, obtained as a result of the previous machining operation, and the desired machining shape. It is also a correction value for correcting the drive trajectory calculated by the drive trajectory calculation unit 311.Therefore, the command value of the axis motion trajectory, which is calculated by the calculation device 312, is also referred to as the correction value of the drive trajectory.

[0075] To estimate the inter-electrode distance, a computational model is obtained beforehand, which describes the relationship between at least one element of data selected from the mean inter-electrode voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation, and the machining shape of the workpiece 11. Using a function that incorporates this computational model, the command value of the axis movement trajectory to achieve the desired shape is then calculated.Regarding the calculation model, the accuracy of the estimated inter-electrode distance increases with an increasing number of data elements selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation. It is therefore preferable to use as many types of data as possible from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory during the previous machining operation. The drive trajectory control unit 31b outputs the calculated drive trajectory command value to the drive control unit 35.

[0076] The axis movement trajectory with respect to the estimated inter-electrode distance is calculated in the normal vector direction, as described above. For example, if the corresponding position protrudes as a result of previous machining (i.e., if the inter-electrode distance is small), the axis movement trajectory is calculated in the direction of workpiece 11 to eliminate the protruding shape. Furthermore, if the corresponding position is recessed as a result of previous machining (i.e., if the inter-electrode distance is large), the correction value of the drive trajectory is calculated in a direction away from workpiece 11 to eliminate the recessed shape.

[0077] When controlling the machine processing speed for a straight shape, the axis movement trajectory ΔL in the normal direction with respect to the axis progress direction is calculated using formula (9) below. Here, C4 is the conversion coefficient for the correction value of the machine processing speed into the axis movement trajectory. ΔL=C4*∫f(Vn−1,Fn−1,Sn−1)dt

[0078] Even when controlling the machine processing speed for a curved shape, the difference between the machine processing shape of the corner section and the desired shape, estimated as a result of previous machining, can be eliminated by controlling the axis motion trajectory. Specifically, the computational device 312 calculates the correction value of the axis motion trajectory at the position of the corner section from the corner section information stored in the drive trajectory memory unit 39, in addition to the profiles of the previous machining, which are stored in the interelectrode mean voltage memory unit 36, the frequency electrical discharge memory unit 37, and the machine processing speed memory unit 38. As described above, the corner section information includes the corner diameter, the opening angle of the corner section, and the like.A formula for calculating the post-correction axis movement trajectory ΔL. comp_cnr The corner section is expressed by formula (10) below. Here, the conversion coefficient for the correction value of the machine processing speed into the axis movement trajectory is represented by C5. ΔLcomp_cnr=C5*∫h(R,θ,Vn−1,Fn−1,Sn−1,x)dt

[0079] As in Fig. 6 and Fig. As shown in Figure 7, if the shape of the corner section is smaller than the desired shape as a result of previous machining, or if there is an unevenness in the entrance / exit of the corner section, it is possible to improve the shape accuracy after machining by calculating the command value of the drive trajectory according to these shapes.

[0080] In the third embodiment, at least one element is stored consisting of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the (n-1)th machining operation of n machining operations in the machining section, where n is an integer of two or more. The inter-electrode distance of the nth machining operation is calculated using a computational model that specifies the relationship between the data of the (n-1)th machining operation and the machining shape of the workpiece 11. The axis movement trajectory of the nth machining operation, which corresponds to the machining volume required to achieve the desired shape, is calculated from the inter-electrode distance of the nth machining operation.This means that the machining shape of workpiece 11 can be estimated with high accuracy during the (n-1)th machining operation, which is the previous machining operation. Therefore, the machining operation can be carried out in such a way that the workpiece 11 has the desired shape during the shape correction machining operation. Accordingly, it is possible to carry out shape correction machining with high accuracy even if a sudden change in shape occurs as a result of the previous machining operation. Fourth embodiment.

[0081] In the first to third embodiments, a calculation model is used when performing a shape correction machine operation, which pre-calculates the inter-electrode distance of the current machine operation at the position from at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine operation speed and the drive trajectory of the previous machine operation.However, all data of the first to (n-1)th rough machining operations can be stored in the inter-electrode mean voltage storage unit 36, the frequency electric discharge storage unit 37, the machining speed storage unit 38 and the drive trajectory storage unit 39, where n is an integer of two or more and the current machining operation, which is a shape correction machining operation, is the nth machining operation, so that the computational devices 342, 222 and 312 can estimate the inter-electrode distance of the current machining operation at the corresponding position using all data of the first to (n-1)th rough machining operations.This makes it possible to stably estimate the inter-electrode distance even if a disturbance impairs the data for the inter-electrode mean voltage, the frequency of electrical discharges, and the machine processing speed at the corresponding location of the previous machining operation. This means that the accuracy of the machining geometry of workpiece 11, obtained as a result of the first to (n-1)th machining operations, is improved compared to using only the data from the (n-1)th machining operation.

[0082] In the above description, at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed and the drive trajectory of the first to (n-1)th machine processing is used; however, at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed and the drive trajectory of two or more of the (n-1)th and earlier machine processing operations may be used.

[0083] In the fourth embodiment, at least one element of data, selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of several machine operations prior to the (n-1)th machine operation, is stored in the inter-electrode mean voltage storage unit 36, the frequency of electrical discharges storage unit 37, the machine processing speed storage unit 38, and the drive trajectory storage unit 39. The inter-electrode distance of the current machine operation is estimated using several elements of data from the (n-1)th and previous machine operations. At least one command value, selected from the machine processing speed, the frequency of electrical discharges, and the drive trajectory, is calculated from the inter-electrode distance.This makes it possible to stably estimate the inter-electrode distance even if, due to a disturbance, there is an impairment in the data of the inter-electrode mean voltage, the frequency of electrical discharges and the machine processing speed at the corresponding location of the previous machine processing. Fifth embodiment.

[0084] In the first to fourth embodiments, command values ​​of several correction values, which are selected from the correction value of the machine processing, the correction value of the frequency of electrical discharges and the correction value of the drive trajectory of the previous machine processing, can be calculated to enable more accurate shape correction.

[0085] Fig. Figure 12 is a block diagram showing an exemplary configuration of a wire EDM machine according to the fifth embodiment. It should be noted that components identical to those of the first embodiment are designated by the same reference numerals, their descriptions omitted, and differences compared to the first embodiment are described.

[0086] In the wire EDM machine 1c according to the fifth embodiment, the control unit 30 further comprises a calculation device 40. The calculation device 40 is a combination of the functions of the calculation devices 342, 222, and 312, which are described in the first, second, and third embodiments. In a shape correction machining operation that improves surface roughness and shape accuracy, the calculation device 40, in particular, calculates the inter-electrode distance of the current machining operation using a calculation model that specifies the relationship between at least one element of data selected from the mean inter-electrode voltage, the frequency of electrical discharges, the machining speed, and the drive trajectory of the previous machining operation, and the machining shape of the workpiece 11.Furthermore, the calculating device 40 calculates several correction values ​​from the inter-electrode distance of the current machine operation. These correction values ​​are selected from the correction value of the machine operation speed, the correction value of the frequency of electrical discharges, and the correction value of the drive trajectory of the current machine operation. These correction values ​​correspond to the amount of machine work required to achieve the desired shape, which is based on the machine operation program for machining the machine operation section. The calculating device 40 outputs the calculation result to the corresponding unit of a machine operation speed control unit 34c, the machine operation power supply control unit 22, and the drive trajectory control unit 31.Accordingly, the command value is corrected by at least two processing units of the machine processing speed control unit 34c, the machine processing power supply control unit 22, and the drive trajectory control unit 31. Alternatively, the computational device 40 can optionally use the corrected command value of the machine processing speed, the frequency of electrical discharges, or the drive trajectory according to the pre-calculation result of the shape or the inter-electrode distance using the computational model.

[0087] The wire EDM machine 1c according to the fifth embodiment comprises the machine processing speed control unit 34c instead of the machine processing speed control unit 34. The machine processing speed control unit 34c corresponds to the machine processing speed calculation unit 341 described in the first embodiment and does not include the calculation device 342. This means that the machine processing speed control unit 34c calculates the machine processing speed in such a way that the difference calculated by the voltage calculation unit 33 becomes zero.

[0088] In the description above, the calculating device 40 performs a calculation using at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the previous machine operation. However, as in the fourth embodiment, a calculation can be performed using at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory in the previous and earlier machine operations.

[0089] In the fifth embodiment, the inter-electrode distance is calculated using a computational model derived from at least one element of data selected from the inter-electrode mean voltage, the frequency of electrical discharges, the machine processing speed, and the drive trajectory of the previous machining operation. Several correction values, selected from the correction value of the machine processing speed, the correction value of the frequency of electrical discharges, and the correction value of the drive trajectory of the current machining operation, are then calculated from the inter-electrode distance. Accordingly, the accuracy of machining the workpiece 11 into the desired shape can be improved compared to the first through fourth embodiments.Furthermore, the question of whether it is better to control the machine processing speed, the frequency of electrical discharges, or the axis movement trajectory can vary depending on the estimated result of the machine processing geometry or the inter-electrode distance of the previous machining of workpiece 11, which is estimated by the computational model. In such a case, the computational device 40 can select the control method, which is determined depending on the estimated result of the machine processing geometry or the inter-electrode distance. Sixth embodiment.

[0090] In the description of the first to fifth embodiments, the inter-electrode mean voltage, the machine processing speed, the frequency of electrical discharges, and the drive trajectory of the previous machine processing or of previous and earlier machine processing operations are stored, and at least one correction value, selected from the machine processing speed, the frequency of electrical discharges, and the drive trajectory of the current machine processing operation, is calculated from these results. The calculation of the correction values ​​for the machine processing speed, the frequency of electrical discharges, and the drive trajectory from the stored profiles is performed using a formula that represents the modeled relationship between the inter-electrode distance or the machining geometry of the workpiece 11 and the calculation objective.In practice, however, the correction value of the calculation target cannot be optimally calculated due to, for example, a deflection of the machining electrode 10 caused by the distance between the upper and lower guides at the time of machining, the thickness of the workpiece 11, the machining fluid pressure, or similar factors, or the occurrence of a reverse electrical discharge in the electrode advance direction, which occurs at a small corner. The calculation target is at least a correction value selected from the machining speed, the frequency of electrical discharges, and the drive trajectory, and which is calculated from the estimated inter-electrode distance.Therefore, the sixth embodiment describes a method for optimizing a target value, which is a value to be calculated, by machine learning from the relationship between the profiles of the inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory, which are obtained by prior machine processing, and the machine processing result of the workpiece 11, that is, the measurement result of the machine processing shape of the workpiece 11 after machine processing.

[0091] The following section describes, sequentially, the learning phase for generating a learned model and the application phase for estimating the command value information of the computation target using the learned model generated in the learning phase. Learning phase

[0092] First, a previous machining operation is described as the learning phase. Fig. Figure 13 is a model diagram illustrating an example of the overview of machine learning according to the sixth embodiment. During the preceding machining operation, training data is collected and stored. Specifically, during the preceding machining operation, the machining specification values ​​of the workpiece 11, the mean interelectrode voltage, machining speed, frequency of electrical discharges, drive trajectory, and instruction value information of the computational target are obtained for each machining operation, along with the measurement results. The relationship between the machining specification values, the mean interelectrode voltage, machining speed, frequency of electrical discharges, drive trajectory, and instruction value information of the computational target for each machining operation, and the measurement result, is generated as a database.Examples of machining specification values ​​for workpiece 11 include the workpiece 11 plate thickness, the workpiece 11 material, the positions of the upper and lower guides, and the wire diameter, which is the diameter of the machining electrode 10. The command value information of the calculation target is at least one element of data selected from the machining speed correction value, the electrical discharge frequency correction value, and the drive trajectory correction value, and is data actually used in the preceding machining operation. The measurement result represents the error of the machining shape of workpiece 11 after machining compared to the desired machining shape as a numerical value.

[0093] Details of the learning phase are described below. Fig. Figure 14 is a diagram schematically showing an exemplary configuration of a learning device used in the control unit for the wire EDM machine according to the sixth embodiment. The learning device 50 comprises a data reference unit 51, a training data storage unit 52, a model generation unit 53, and a learned model storage unit 54.

[0094] The data reference unit 51 acquires training data. The training data comprises the machining specification values ​​of the workpiece 11 from the preceding machining operation, the inter-electrode mean voltage, machining speed, frequency of electrical discharges, drive trajectory, and command value information of the computation target for each machining operation, as well as the measurement result. The data reference unit 51 outputs the machining specification values, the inter-electrode mean voltage, the machining speed, the frequency of electrical discharges, the drive trajectory, and the command value information of the computation target from the acquired training data as learning data to the model generation unit 53. The data reference unit 51 stores the acquired training data in the training data storage unit 52.

[0095] It is preferred that the inter-electrode mean voltage, the machine processing speed, the frequency of electrical discharges, and the drive trajectory are all included; however, at least one of these data points should be included. The following description presents an example of a case in which the inter-electrode mean voltage, the machine processing speed, the frequency of electrical discharges, and the drive trajectory are all included.

[0096] The training data storage unit 52 stores, as a database, the relationship between the referenced training data, i.e., the machine processing specification values ​​of the workpiece 11, the inter-electrode mean voltage, machine processing speed, frequency of electrical discharges, drive trajectory, and command value information of the calculation target for each machine processing operation, and the measurement result. The training data stored in the training data storage unit 52 is used in the application phase, which is described later.

[0097] The model generation unit 53 learns an initial computation target instruction value, which is an instruction value of the computation target, based on the training data. This training data is generated from the combination of the machine processing specification values, the inter-electrode mean voltage, the machine processing speed, the frequency of electrical discharges, and the drive trajectory output by the data reference unit 51, and the computation target instruction value information, which is response data. This means that the learned model is generated to infer the optimal initial computation target instruction value from the machine processing specification values, the inter-electrode mean voltage, the machine processing speed, the frequency of electrical discharges, the drive trajectory, and the computation target instruction value information.The training data consists of machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency, and drive trajectory associated with the instruction value information of the computation target. The training data is a combination of the machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency, drive trajectory, and the instruction value information of the computation target. However, it can also be a combination of the machine processing specification values, at least one element of data selected from the inter-electrode mean voltage, machine processing speed, discharge frequency, and drive trajectory, and the instruction value information of the computation target.It should be noted that the command value information of the computation target, which is used when generating the learned model, consists of data that expresses the desired form as a command value.

[0098] The learning algorithm used by the model generation unit 53 can be a well-known algorithm, such as supervised learning. As an example, a case is described in which a neural network is used.

[0099] According to one example, the model generation unit 53 learns the first computation target command value using what is called supervised learning based on a neural network. In this context, supervised learning refers to a model that provides the learning unit 50 with pairs of inputs and labels, which are outputs, in order to learn features in this training data and to infer outputs from inputs.

[0100] The neural network comprises an input layer consisting of several neurons, an intermediate layer consisting of several neurons, and an output layer consisting of several neurons. The number of intermediate layers, also called hidden layers, can be one, two, or more.

[0101] Fig. Figure 15 is a diagram that schematically shows an example of a neural network used by the model generation unit. According to an example, in the case of a three-layer neural network, as described in Fig. As shown in Figure 15, multiple inputs are fed into input layers X1 to X3, and these values ​​are multiplied by weights represented by w11 to w16 and fed into intermediate layers Y1 to Y2. Weights w11 to w16 are referred to as weights w1 unless they are individually distinguished. Furthermore, the results from intermediate layers Y1 and Y2 are multiplied by weights represented by w21 to w26 and output from output layers Z1 to Z3. Weights w21 to w26 are referred to as weights w2 unless they are individually distinguished. The output results of output layers Z1 to Z3 vary depending on the values ​​of weights w1 and w2.

[0102] In the sixth embodiment, the neural network learns the first computation target instruction value through what is called supervised learning, based on the learning data generated from the combination of the machine processing specification values, the inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory obtained by the data reference unit 51, and the instruction value information of the computation target.

[0103] This means that the neural network learns by adjusting the weights w1 and w2 so that the result, which is output from the output layer in response to the input of the machine processing specification values, inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory into the input layer, approximates the instruction value information of the computation target.

[0104] The model generation unit 53 performs the learning process as described above in order to generate and output a learned model.

[0105] The learned model storage unit 54 stores the learned model, which is output from the model generation unit 53.

[0106] Next, a learning process using the learning device 50 will be described. Fig. Figure 16 is a flowchart illustrating an exemplary procedure for learning processing by the learning device according to the sixth embodiment. First, the data reference unit 51 obtains the machine processing specification values, inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory, the command value information of the computation target, which are response data, and the measurement result (step S51). The machine processing specification values, inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory, the command value information of the computation target, and the measurement result are obtained simultaneously.However, the machine processing specification values, inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory, the instruction value information of the calculation target and the measurement result only need to be entered in association with each other, and data of the machine processing specification values, inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory, the instruction value information of the calculation target and the measurement result can be obtained at different times.

[0107] Next, the training data storage unit 52 stores the machine processing specification values, inter-electrode mean voltage, machine processing speed, frequency of electrical discharges and drive trajectory, the instruction value information of the computation target and the measurement result (step S52).

[0108] The model generation unit 53 then learns the first computation target instruction value through what is called supervised learning, based on the learning data generated from the combination of the machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency and drive trajectory obtained by the data reference unit 51, and the computation target instruction value information, and generates a learned model (step S53).

[0109] The learned model storage unit 54 stores the learned model, which is generated by the model generation unit 53 (step S54). Processing then ends. Application phase

[0110] Next, the application phase, a machining process, is described. Fig. Figure 17 is a model diagram illustrating an example of the overview of machine learning according to the sixth embodiment. During machining, a target value is calculated by machine learning to minimize the error relative to the desired shape. The profiles of the previous machining operation, or of previous and earlier machining operations, are used as input data, the database as training data, and the first and second computation target command values ​​(which are command values ​​of computation targets) as output data. This allows the formula representing the modeled relationship between the inter-electrode distance and the computation target, as well as the error of actual machining, to be compensated for by machine learning.Any algorithm can be used as a model for such machine learning, for example, neighbor regression or Bayesian optimization.

[0111] Fig. Figure 18 is a diagram schematically showing an exemplary configuration of a tracking device used in the control unit for the wire EDM machine according to the sixth embodiment. The tracking device 60 comprises a data reference unit 61 and a tracking unit 62.

[0112] Data reference unit 61 retrieves the machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency, and drive trajectory. For example, data reference unit 61 retrieves the machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency, and drive trajectory of the previous (n-1)th machine processing operation.

[0113] The inference unit 62 infers the first computation target command value, which is obtained using the learned model. This means that by inputting the machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency, and drive trajectory, which are obtained by the data reference unit 61, into the learned model, the first computation target command value can be inferred from the machine processing specification values, the inter-electrode mean voltage, the machine processing speed, the electrical discharge frequency, and the drive trajectory. In the described example, the first computation target command value is inferred from the machine processing specification values, the inter-electrode mean voltage, the machine processing speed, the electrical discharge frequency, and the drive trajectory.However, the first computation target command value can be inferred from the machining specification values ​​and at least one element of data selected from the inter-electrode mean voltage, the machining speed, the electrical discharge frequency, and the drive trajectory. In this case, a learned model is used to infer the first computation target command value from the machining specification values ​​and at least one element of data selected from the inter-electrode mean voltage, the machining speed, the electrical discharge frequency, and the drive trajectory. When inferring the first computation target command value, the input data does not include a measurement result.This is because there is no need to intentionally create an irregularity in the shape obtained as a result of machine processing, and a command value that produces the smallest shape variation is inferred.

[0114] Based on the inference of the first computation target command value using the learned model, the inference unit 62 calculates the second computation target command value, which is a command value of a different computation target that minimizes the error compared to the desired shape when the correction value of the computation target is set to the first computation target command value, using the training data in the training data storage unit 52. The second computation target command value of a different computation target is the command value of at least one remaining computation target, excluding the computation target that is the first computation target command value, from the correction value of the machine processing speed, the correction value of the frequency of electrical discharges, and the correction value of the drive trajectory.The training data storage unit 52 stores as a database the relationship between the machine processing specification values ​​of the workpiece 11, the inter-electrode mean voltage, machine processing speed, frequency of electrical discharges, drive trajectory and command value information of the calculation target for each machine processing operation and the measurement result.In the sixth embodiment, after estimating the first computation target instruction value using the learned model, the inference unit 62 therefore calculates a change in the error associated with a change in the instruction value of another computation target, excluding the first computation target instruction value, for example, the machine processing speed, the frequency of electrical discharges, or the drive trajectory, using the training data stored in the training data storage unit 52, and obtains and outputs the second computation target instruction value of the computation target, which minimizes the error.This means that the inference unit 62, using the training data in the training data storage unit 52, learns and calculates the second computation target instruction value of another computation target, which minimizes the error from the desired shape when the correction value of the computation target is set to the first computation target instruction value.

[0115] In the description of the sixth embodiment, the first calculation target command value is inferred using the learned model, which was learned by the model generation unit 53 of the control unit for the wire EDM machine 1, and the second calculation target command value of a different calculation target, which minimizes the error when the correction value of the calculation target is set to the first calculation target command value, is output. However, a learned model can be obtained externally, for example, from the control unit of another wire EDM machine 1, the first calculation target command value can be inferred based on the learned model, and the second calculation target command value of a different calculation target, which minimizes the error when the correction value of the calculation target is set to the first calculation target command value, can be output.

[0116] Next, inference processing by the inference device 60 is described. Fig. Figure 19 is a flowchart illustrating an exemplary procedure for inference processing by the inference device according to the sixth embodiment. First, the data reference unit 61 obtains the machine processing specification values, the inter-electrode mean voltage, the machine processing speed, the frequency of electrical discharges, and the drive trajectory (step S71).

[0117] Next, the inference unit 62 inputs the machine processing specification values, the inter-electrode mean voltage, the machine processing speed, the frequency of electrical discharges, and the drive trajectory into the learned model, which is stored in the learned model memory unit 54, and infers the first computation target instruction value (step S72). Then, using the training data in the training data memory unit 52, the inference unit 62 calculates the second computation target instruction value of a different computation target, which minimizes the error if the first computation target instruction value obtained by the learned model is set as the computation target correction value (step S73).

[0118] The inference unit 62 then outputs the calculated first calculation target command value and the second calculation target command value to the corresponding processing units in the machine processing speed control unit 34c, the machine processing power supply control unit 22 and the drive trajectory control unit 31 (step S74).

[0119] Each processing unit then uses the output first computation target command value and second computation target command value to correct the command value calculated by the respective processing unit, namely the machine processing speed, the frequency of electrical discharges, and the drive trajectory (step S75). This enables machining to be carried out according to the desired shape.

[0120] It should be noted that the model generation unit 53 can learn the first computation target instruction value based on the training data generated for the control units for multiple wire EDM machines 1. It should also be noted that the model generation unit 53 can obtain training data from the control units for multiple wire EDM machines 1 used in the same area, or it can learn the first computation target instruction value using training data collected from the control units for multiple wire EDM machines 1 operating independently in different areas. Furthermore, during the learning process, it is possible to start collecting training data from the control unit for a new wire EDM machine 1, or to stop collecting training data from the control unit for an existing wire EDM machine.Furthermore, the learning device 50, which has learned the first calculation target command value for the control unit for a specific wire EDM machine 1, can be applied to the control unit of another wire EDM machine 1, and the first calculation target command value can be relearned and updated for the control unit of the other wire EDM machine 1.

[0121] Furthermore, the learning device 50 and the inference device 60, which are used to learn the first calculation target command value of the control unit for the wire EDM machine 1, can, for example, be a device that is separate from the control unit for the wire EDM machine 1 and connected to it via a network. Alternatively, the learning device 50 and the inference device 60 can be integrated into the control unit for the wire EDM machine 1. Finally, the learning device 50 and the inference device 60 can exist on a cloud server.

[0122] In the sixth embodiment, the first computation target command value is learned based on training data generated from a combination of machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency, drive trajectory, and the computation target command value information (response data). A learned model for inferring the first computation target command value is then generated. Next, the machine processing specification values, inter-electrode mean voltage, machine processing speed, electrical discharge frequency, and drive trajectory of the previous machine processing operation are fed into the learned model, and the first computation target command value of the current machine processing operation is inferred.If the correction value of the calculation target is set to the first calculation target command value, the second calculation target command value, which is the command value of another calculation target that minimizes the error compared to the desired shape, is calculated using training data that includes the machine machining specification values, the inter-electrode mean voltage, the machine machining speed, the frequency of electrical discharges, the drive trajectory, and the command value information of the calculation target of each machine machining operation and the measurement result, which are stored as a database.Using a formula that represents the modeled relationship between the inter-electrode distance and the calculation target as in the first to fifth embodiments, the value cannot be optimally calculated in practice due to, for example, deflection of the electrode due to the distance between the upper and lower guide during machining, the thickness of the workpiece 11, the machining fluid pressure or the like, or the occurrence of a backward electrical discharge in the electrode advance direction, which occurs at a small corner.Even if a deflection of the electrode occurs due to the distance between the upper and lower guides during machining, the thickness of the workpiece 11, the machining fluid pressure, or the like, or a backward electrical discharge in the electrode advance direction occurring at a small corner, or the like, the target value can be optimized in the sixth embodiment by machine learning from the relationship between the profiles of the inter-electrode mean voltage, the machining speed, the frequency of electrical discharges, and the drive trajectory obtained from previous machining, and the measurement result.

[0123] The machine processing power supply control unit 22 and the control unit 30, which are described in the first to sixth embodiments, correspond to the control unit for the wire EDM machine 1. Next, a hardware configuration for implementing the control unit is described. The control unit is implemented by a processing circuit, which is a circuit in which a processor executes software. According to an example, the processing circuit, which executes software that is in Fig. 20 control circuits shown. Fig. Figure 20 is a diagram showing an exemplary hardware configuration of the control unit for the wire EDM machine according to the first to sixth embodiments. The control circuit 100 comprises an input unit 101, a processor 102, a memory 103, and an output unit 104.

[0124] The input unit 101 is an interface circuit that receives data input from outside the control circuit 100 and provides data to the processor 102. The output unit 104 is an interface circuit that transmits data from the processor 102 or the memory 103 to an external location outside the control circuit 100. In a case where the processing circuit is in Fig. In the control circuit 100 shown in Figure 20, the processor 102 reads the program, which corresponds to each of the components of the machine processing power supply control unit 22 and the control unit 30 and is stored in memory 103, and executes it, thereby implementing each of the components. Memory 103 is also used as temporary storage for each process carried out by the processor 102. The processor 102 can output data, such as calculation results, to memory 103 and cause memory 103 to store the data, or it can cause an auxiliary storage device to store data, such as calculation results, via the volatile memory of memory 103.

[0125] The processor 102 is a central processing unit (CPU, also referred to as a central processing unit, a machine processing unit, a computing unit, a microprocessor, a microcomputer, a processor, or a digital signal processor (DSP)). Examples of memory 103 include non-volatile or volatile semiconductor memory, a magnetic floppy disk, a flexible floppy disk, an optical disk, a compact disk, a mini disk, a digital versatile disk (DVD), and the like. Examples of non-volatile or volatile semiconductor memory include random-access memory (RAM), read-only memory (ROM), flash memory, EPROM (erasable programmable read-only memory), EEPROM (registered trademark) (electrically erasable programmable read-only memory), and the like.

[0126] Fig.Figure 20 is an example of hardware for implementing each of the components of the machine processing power supply control unit 22 and the control unit 30 with the general-purpose processor 102 and the memory 103; however, each of the components can be implemented by a dedicated hardware circuit. The processing circuit, which is a dedicated hardware circuit, is a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an application-specific integrated circuit (ASIC), an FPGA (field programmable gate array), or a circuit comprising a combination of these. Each of the components can be implemented by a combination of the control circuit 100 and a dedicated hardware circuit.

[0127] The configurations described in the embodiments mentioned above are examples. The embodiments can be combined with other well-known techniques and with each other, and some of the configurations can be omitted or modified to an extent that does not deviate from the basic idea. Reference symbol list 1, 1a, 1b, 1c wire spark erosion machine; 10 Machining electrode; 11 workpiece; 20 power supply units; 21 Machine processing power supply; 22, 22a Machine processing power supply control unit; 30 control unit; 31, 31b Drive trajectory control unit; 32 Inter-electrode mean voltage detection unit; 33 Voltage calculation unit; 34, 34a, 34b, 34c Machine processing speed control unit; 35 Drive control unit; 36 Inter-electrode medium-voltage storage unit; 37 Frequency-Electrical Discharge Storage Unit; 38 machine processing speed storage unit; 39 Drive trajectory storage unit; 40, 222, 312, 342 Calculation device; 50 learning devices; 51, 61 Data reference unit; 52 training data storage unit; 53 Model generation unit; 54 Learned Model Memory Unit; 60 Inference device; 62 Inference unit; 221 Frequency-Electrical Discharge Control Unit; 311 Drive trajectory calculation unit; 341 Machine processing speed calculation unit.

Claims

[1] Control unit (20, 30) for a wire EDM machine (1; 1a; 1b; 1c) which machine a workpiece (11) by applying a voltage between the workpiece (11) and an electrode (10) and causing an electrical discharge, wherein the control unit (20, 30) is configured to control a drive trajectory of the electrode (10) relative to the workpiece (11), a relative machine machining speed between the workpiece (11) and the electrode (10) and a frequency of electrical discharges of a voltage periodically applied between the electrode (10) and the workpiece (11), wherein the control unit (20, 30) comprises: a storage device (36, 37, 38, 39) which stores at least one element of data selected from a voltage mean value, which is a mean value of a voltage applied between the workpiece (11) and the electrode (10), the frequency of electrical discharges, the machine machining speed and the drive trajectory of an (n-1)th machine machining operation of n machine machining operations of a predetermined machine machining section of the workpiece (11), where n is an integer of two or more; and a calculating device (342) which calculates an electrode gap, which is a distance between the workpiece (11) and the electrode (10), of an nth machining operation using a calculation model which specifies a relationship between the data of the (n-1)th machining operation and a machining shape of the workpiece (11), and which calculates from the electrode gap of the nth machining operation at least one command value, which is selected from the machining speed, the frequency of electrical discharges and the drive trajectory of the nth machining operation, which corresponds to a machining quantity required to achieve a desired shape, which is based on a machining program for machining the machining section. [2] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 1, wherein the storage device (36, 37, 38, 39) furthermore stores at least one element of the data selected from the voltage mean, the frequency of electrical discharges, the machine processing speed and the drive trajectory of a machine processing operation prior to the (n-1)th machine processing operation, and the calculating device (342) calculates at least one command value, which is selected from the machine processing speed, the frequency of electrical discharges and the drive trajectory of the nth machine processing, using the calculation model which specifies a relationship between the data of the (n-1)th and previous machine processing and the machine processing shape of the workpiece (11). [3] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 1 or 2, further comprising a command value calculation unit (311) which calculates at least one command value of the machine processing speed, the frequency of electrical discharges and the drive trajectory using at least one element from the voltage mean, the frequency of electrical discharges and the drive trajectory of the nth machine processing operation and which corrects the calculated command value with the at least one command value calculated by the calculation device (342) which is selected from the machine processing speed, the frequency of electrical discharges and the drive trajectory of the nth machine processing operation. [4] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 1, wherein, if the machine machining section has a straight shape, the calculation model includes a function which specifies a relationship between at least one element of the data selected from the voltage mean, the frequency of electrical discharges, the machine machining speed and the drive trajectory of the (n-1)th machine machining, and the machine machining shape of the workpiece (11). [5] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 2, wherein, if the machine machining section has a straight shape, the calculation model includes a function which specifies a relationship between at least one element of the data selected from the voltage mean, the frequency of electrical discharges, the machine machining speed and the drive trajectory of the (n-1)th and previous machine machining operations, and the machine machining shape of the workpiece (11). [6] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 1, wherein, if the machine machining section is a corner section with a curved shape, the calculation model includes a function which specifies a relationship between a corner diameter and an opening angle of the corner section, at least one element of the data selected from the stress mean, the frequency of electrical discharges, the machine machining speed and the drive trajectory of the (n-1)th machine machining, and a shape of the workpiece (11). [7] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 2, wherein, if the machine machining section is a corner section with a curved shape, the calculation model includes a function which specifies a relationship between a corner diameter and an opening angle of the corner section, at least one element of the data selected from the stress mean, the frequency of electrical discharges, the machine machining speed and the drive trajectory of the (n-1)th and previous machine machining operations, and a shape of the workpiece (11). [8] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 1, further comprising a learning device (50) which includes: a data reference unit (51) which receives training data comprising machining specification values ​​of the workpiece (11) of a previous machining operation, at least one element of data selected from the stress mean, machining speed, electrical discharge frequency and drive trajectory of each machining operation, and computation value information which is a target of the command value; and a model generation unit (53) which generates a learned model for inferring a first computation target instruction value, which is a target of the instruction value, from the machine processing specification values ​​and at least one element of data selected from the stress mean, machine processing speed, frequency of electrical discharges and drive trajectory, which are obtained from the data reference unit (51). [9] Control unit (20, 30) for the wire EDM machine (1; 1a; 1b; 1c) according to claim 1, further comprising a follower device (60) which includes: a training data storage unit (52) which stores training data which specifies a relationship between machining specification values ​​of the workpiece (11) of a previous machining operation, the stress mean value, the machining speed, the frequency of electrical discharges, the drive trajectory and a calculation value information which is a target of the command value, of each of the machining operations and which specifies a measurement result which indicates a deviation of a machining shape from a desired shape in each of the machining operations; a data reference unit (61) which references machining specification values ​​of the workpiece (11) and at least one element of data selected from the stress mean, machining speed, electrical discharge frequency and drive trajectory of the (n-1)th machining operation; and a inference unit (62) which, using a learned model for inferring a first computation target instruction value, which is an instruction value of a computation target of the nth machine operation, from the machine operation specification values ​​and at least one element of data selected from the stress mean, machine operation speed, frequency of electrical discharges and drive trajectory of the (n-1)th machine operation, infers the first computation target instruction value from the machine operation specification values ​​and at least one element of data selected from the stress mean, machine operation speed, frequency of electrical discharges and drive trajectory of the (n-1)th machine operation, which are obtained by the data reference unit (61), and which, using the training data, infers a second computation target instruction value,which is a command value of another computation goal, calculates which minimizes an error of a desired shape when the command value of the computation goal is set to the first computation goal command value. [10] Control method for a wire EDM machine (1; 1a; 1b; 1c) which machine a workpiece (11) by applying a voltage between the workpiece (11) and an electrode (10) and causing an electrical discharge, wherein the control method is used to control a drive trajectory of the electrode (10) relative to the workpiece (11), a relative machine machining speed between the workpiece (11) and the electrode (10) and a frequency of electrical discharges of a voltage periodically applied between the electrode (10) and the workpiece (11), wherein the control method comprises: a storage step in which a storage device (36, 37, 38, 39) stores at least one element of data selected from a voltage mean value, which is a mean value of a voltage applied between the workpiece (11) and the electrode (10), the frequency of electrical discharges, the machine processing speed and the drive trajectory of an (n-1)th machine operation of n machine operations of a predetermined machine processing section of the workpiece (11), where n is an integer of two or more; and a calculation step in which a calculating device (342) calculates an electrode distance, which is a distance between the workpiece (11) and the electrode (10), of an nth machine operation using a calculation model which specifies a relationship between the data of the (n-1)th machine operation and a machine operation shape of the workpiece (11), and from the electrode distance of the nth machine operation calculates at least one command value, which is selected from the machine operation speed, the frequency of electrical discharges and the drive trajectory of the nth machine operation, which corresponds to a machine operation quantity required to achieve a desired shape, which is based on a machine operation program for machine operation of the machine operation section.

Images