Multi-wire electrical discharge machining (EDM) machine and multi-wire EDM machining system

Abstract

Multi-wire electrical discharge machining (EDM) machine (1) comprising a cutting wire section (6a) which includes several cutting wires arranged in parallel and facing a workpiece (2), wherein the several cutting wires extend parallel between several guide rollers (3c, 3d) by having a single wire electrode (6) wound around the several guide rollers (3c, 3d), and which cuts the workpiece (2) by causing electrical discharge machining between the cutting wire section (6a) and the workpiece (2), wherein the multi-wire electrical discharge machining (1) comprises: a machining power source (7) which applies a machining pulse voltage between the workpiece (2) and the cutting wire section (6a);and a wire break detection unit (20) which detects a break in the wire electrode (6), wherein the wire break detection unit (20) is electrically connected in parallel to a wire electrode section which extends over a region from a location on a feed side of the wire electrode (6) to a location on an ejection side of the wire electrode (6), and the wire break detection unit (20) comprises: a wire break detection circuit (206) which is an electrical circuit in which a wire break detection energy source (204) and a current-limiting resistor (205) are connected in series, wherein the wire break detection energy source (204) supplies a constant direct current to the wire electrode section at least during a cutting process on the workpiece (2); and a wire break determination unit (207) which determines a break in the wire electrode (6) based on a current flowing through the current-limiting resistor (205).

Description

Area

[0001] The present disclosure relates to a multi-wire electrical discharge machining (EDM) machine and in particular a multi-wire EDM machine comprising a cutting wire section which includes several wire electrode sections (hereinafter referred to as cutting wires) which are arranged in parallel and positioned facing a workpiece, wherein the several cutting wires extend parallel between several guide rollers by having a single wire electrode wound around the several guide rollers, and which cuts the workpiece into several pieces by applying a pulse voltage which causes EDM between the cutting wire section and the workpiece. background

[0002] As a means of cutting a wafer-like thin plate from a columnar workpiece, such as a semiconductor ingot, a multi-wire electrical discharge machining (EDM) machining technique has already been disclosed, in which a cutting wire section comprising several cutting wires is formed for a columnar workpiece using a single wire electrode wound around several guide rollers, and in which a pulse voltage is applied independently to cause EDM between the cutting wire section and the columnar workpiece, so that several thin plates are cut together from the workpiece (see, for example, patent literature 1).

[0003] In electrical machining, such as multi-wire electrical discharge machining (EDM), a wire electrode break (hereinafter referred to as wire breakage) necessitates an immediate halt to the machining process to prevent damage to the workpiece or machine. Therefore, rapid and accurate wire electrode breakage detection is essential. Patent literature 1 proposes a method for detecting a wire electrode breakage by applying a pulse voltage to the electrode only during a period when no pulse voltage is applied by a machining power source, and by examining the conductivity of the wire electrode. Citation list of patent literature

[0004] Patent literature 1: Japanese patent application, publication number 2009-184071 Brief description: Technical problem

[0005] A conventional multi-wire electrical discharge machining (EDM) machine is configured such that a single wire electrode is wound around several guide rollers. As the number of turns increases, the inductance of the wire electrode also increases, making it difficult to induce a high-frequency current to flow through it. Accordingly, the method disclosed in patent literature 1, in which a pulse voltage is applied to a wire electrode for break detection and the conductivity of the wire electrode is checked, has the problem that a wire break cannot be detected as the number of turns of the wire electrode increases.

[0006] The present disclosure was made in consideration of the above, and one objective of the present disclosure is to provide a multi-wire electrical discharge machining (EDM) machine capable of detecting a wire break quickly and accurately. Solution to the problem

[0007] To solve the problem described above and achieve the objective, a multi-wire electrical discharge machining (EDM) machine of the present disclosure comprises a cutting wire section containing several cutting wires arranged in parallel and facing a workpiece, the multiple cutting wires extending parallel between several guide rollers by means of a single wire electrode wound around the multiple guide rollers. The multi-wire EDM machine cuts the workpiece by generating spark erosion between the cutting wire section and the workpiece. The multi-wire EDM machine comprises: a machine power source that applies a machine pulse voltage between the workpiece and the cutting wire section; and a wire break detection unit that detects a break in the wire electrode.The wire break detection unit is electrically connected in parallel to a wire electrode section extending over a region from a location on a feed side of the wire electrode to a location on an ejection side of the wire electrode, and the wire break detection unit comprises: a wire break detection circuit, which is an electrical circuit in which a wire break detection energy source and a current-limiting resistor are connected in series, wherein the wire break detection energy source supplies a constant direct current to the wire electrode section at least during one execution of a cutting process on the workpiece; and a wire break determination unit, which determines a break in the wire electrode based on a current flowing through the current-limiting resistor. Advantageous effects of the invention

[0008] The multi-wire electrical discharge machine according to the present disclosure enables a wire break to be detected quickly and accurately. List of characters Fig. Figure 1 is a perspective view showing a configuration of a main part of a multi-wire electrical discharge machine according to a first embodiment. Fig. Figure 2 is a diagram which, in the form of an equivalent circuit, shows a relationship between a wire break detection unit and a wire electrode section between a connection target 20a and a connection target 20b of the wire break detection unit during the execution of a spark erosion machine operation by the multi-wire spark erosion machine according to the first embodiment. Fig. Figure 3 is a circuit diagram showing a configuration of the wire break detection unit of the multi-wire electrical discharge machine according to the first embodiment. Fig. Figure 4 is a block diagram showing a hardware configuration of a machine processing control unit according to the first embodiment. Fig. Figure 5 is a circuit diagram showing a configuration of a wire break detection unit of a multi-wire electrical discharge machine according to a second embodiment. Fig. Figure 6 is a circuit diagram showing a configuration of a wire break detection unit of a multi-wire electrical discharge machining (EDM) machine according to a third embodiment. Fig. Figure 7 is a diagram that describes the reaction of a binaryization circuit according to the third embodiment before and after a wire break. Fig. Figure 8 is a diagram that describes the reaction of a low-pass characteristic unit according to the third embodiment before and after a wire break. Fig. Figure 9 is a circuit diagram showing a configuration of a wire break detection unit of a multi-wire electrical discharge machining (EDM) machine according to a fourth embodiment. Fig. Figure 10 is a diagram describing the response of a frequency detection unit according to the fourth embodiment before and after a wire break. Fig. Figure 11 is a perspective view showing a configuration of a main part of a multi-wire electrical discharge machine according to a fifth embodiment. Fig. Figure 12 is a diagram showing a configuration of a machine learning device according to the fifth embodiment. Fig. Figure 13 is a diagram showing an example of a three-layer neural network according to the fifth embodiment. Fig. Figure 14 is a flowchart relating to a learning process to be carried out by the machine learning device according to the fifth embodiment. Fig. Figure 15 is a configuration diagram of a wire break prediction unit relating to the multi-wire electrical discharge machine according to the fifth embodiment. Fig. Figure 16 is a diagram showing a configuration of a multi-wire electrical discharge machining (EDM) system according to the fifth embodiment. Fig. Figure 17 is a diagram showing another configuration of the multi-wire electrical discharge machining system according to the fifth embodiment. Fig. Figure 18 is a flowchart relating to a processing operation to obtain a prediction information using the wire break prediction unit according to the fifth embodiment. Description of embodiments

[0009] Multi-wire electrical discharge machining (EDM) machines according to embodiments of the present disclosure are described in detail below with reference to the drawings. It should be noted that the present disclosure is not limited to these embodiments. First embodiment.

[0010] Fig. Figure 1 is a perspective view showing a configuration of a main part of a multi-wire electrical discharge machine in a first embodiment of the present disclosure.

[0011] A multi-wire electrical discharge machining (EDM) machine 1 according to the first embodiment comprises: a wire spool 4, a wire spool drive unit 17, wire ejection rollers 5, guide rollers 3a, 3b, 3c and 3d, a guide roller drive unit 18, a power supply element 12, a drive unit 14, a machine processing power source 7, a machine processing control unit 9, a wire break detection unit 20 and a machine processing condition detection device 15. The wire spool 4 provides a wire electrode 6. The wire spool drive unit 17 drives the wire spool 4. The wire ejection rollers 5 eject the wire electrode 6 from the multi-wire EDM machine 1. The guide rollers 3a, 3b, 3c and 3d ensure that the wire electrode 6 moves correctly. The guide roller drive unit 18 drives the guide roller 3a.The energy supply element 12 provides a pulse voltage for machining (hereinafter referred to as a machining pulse voltage) to the wire electrode 6 by coming into contact with it. The drive unit 14 moves a machining table 70, on which a workpiece 2 can be placed, in a vertical direction. The machining energy source 7 applies the machining pulse voltage between a cutting wire section 6a and the workpiece 2 via the energy supply element 12. The machining control unit 9 controls the electrical discharge machining (EDM) process. The wire break detection unit 20 detects a break in the wire electrode. The machining status detection unit 15 detects a machining status in the cutting wire section 6a.It should be noted that the multi-wire electrical discharge machine 1 has an orthogonal coordinate system in which a vertical direction is defined as a Z-axis direction and two mutually orthogonal directions in a horizontal plane are defined as an X-axis direction and a Y-axis direction.

[0012] The four guide rollers 3a to 3d are spaced apart from one another and arranged parallel in an axial direction, such that the positions of their axes form a quadrilateral shape in a plane orthogonal to the axial direction. Specifically, the four guide rollers 3a to 3d are arranged as follows: Guide rollers 3a and 3b are positioned at the highest points, guide roller 3c is positioned below guide roller 3b, and guide roller 3d is positioned below guide roller 3a and adjacent to guide roller 3c. Furthermore, numerous guide grooves for guiding the movement of the wire electrode 6 are formed on the outer surfaces of the four guide rollers 3a to 3d at axial intervals.

[0013] In the multi-wire electrical discharge machining (EDM) machine 1 described above, the single wire electrode 6, fed from the wire spool 4, is guided through the guide grooves of the guide rollers 3a, 3b, 3c, and 3d, which rotate clockwise in the example shown. The electrode is wound multiple times around the four guide rollers 3a, 3b, 3c, and 3d at intervals corresponding to the distances between the guide grooves. The wire electrode 6 is then ejected from the multi-wire EDM machine 1 by the wire ejection rollers 5. In this case, a section of the wire electrode 6, extending parallel between guide roller 3c and guide roller 3d, serves as the cutting wire section 6a.

[0014] Furthermore, in the illustrated example, the guide roller 3a is driven by the guide roller drive unit 18, and the wire spool drive unit 17 drives the wire spool 4 such that the voltage of the wire electrode 6 remains constant. The guide roller drive unit 18 and the wire spool drive unit 17 also drive the guide roller 3a and the wire spool 4, respectively, such that a desired wire feed speed is achieved.

[0015] To move the machining table 70 in the Z-axis direction, a position command is issued from the machining control unit 9 to the drive unit 14. This changes the relative position between the workpiece 2 and the cutting wire section 6a. Therefore, the workpiece 2 is moved according to the position command issued by the machining control unit 9 so that it approaches the cutting wire section 6a, and a pulse voltage is applied between the cutting wire section 6a and the workpiece 2 to cause electrical discharge machining (EDM), thus cutting several thin sheets from the workpiece 2. Furthermore, the machining control unit 9 performs feed control in one cutting direction so that a suitable electrode gap is maintained between the workpiece 2 and the cutting wire section 6a (hereinafter referred to as the electrode gap).

[0016] It should be noted that an ingot-like workpiece, which can be cut into several thin sheets, can be used as workpiece 2. Examples of workpiece 2 materials include metals, such as tungsten and molybdenum, which serve as sputtering targets; ceramics, such as polycrystalline silicon carbide, which can be used as various structural elements; semiconductor materials, such as monocrystalline silicon, monocrystalline silicon carbide, monocrystalline gallium nitride, monocrystalline gallium oxide, and monocrystalline diamond, which serve as semiconductor device wafers; and solar cell materials, such as monocrystalline or polycrystalline silicon, which serves as solar cell wafers.

[0017] Among the materials mentioned above, metals can be readily used for electrical discharge machining (EDM). Furthermore, a semiconductor material and a solar cell material suitable for EDM machining can have a sufficiently low resistivity of approximately 100 Ω cm⁻¹ or less, and preferably 10 Ω cm⁻¹ or less. Therefore, a suitable workpiece 2 is a metal or a material with a resistivity equal to or greater than that of metal and 100 Ω cm⁻¹ or less, preferably 10 Ω cm⁻¹ or less. In particular, a semiconductor material and a solar cell material with a resistivity in the range described above are suitable for workpiece 2.

[0018] Then, a machining fluid is introduced into the electrode gap between the workpiece 2 and the cutting wire area 6a by spraying or immersion, as in a so-called single-wire electrical discharge machining (EDM) machine.

[0019] The machining power source 7 generates a machining pulse voltage, which is applied to the electrode gap between the workpiece 2 and the cutting wire area 6a, based on a machining voltage application command from the machining control unit 9. The machining pulse voltage is generated, for example, by a switching power supply system. Furthermore, the machining power source 7 comprises several machining power supply units 8, which can apply voltage independently of one another. A ground electrode 10 is then provided via the several machining power supply units 8 of the machining power source 7. Each ground lead of the several machining power supply units 8 is connected to the ground electrode 10, and the ground electrode 10 is connected to the workpiece 2 by a cable 11.It should be noted that the polarity of the machine processing pulse voltage to be generated can be reversed as required.

[0020] The energy supply element 12 comprises several energy supply element units 13, which are electrically insulated from one another. In the example shown, the wire electrode section extending parallel between the guide rollers 3b and 3c serves as an energy supply wire section 6b, to which the machine processing pulse voltage is supplied. Each energy supply element unit 13 is arranged in sliding contact with a corresponding wire electrode of the energy supply wire section 6b in a one-to-one correspondence. Each energy supply element unit 13 then supplies a machine processing pulse voltage, generated by a corresponding machine processing energy supply unit 8, to each corresponding wire electrode of the energy supply wire section 6b independently.Therefore, the machine processing pulse voltage is applied independently to each cutting wire of the cutting wire area 6a by the respective machine processing power supply unit 8.

[0021] The machine machining condition detection device 15 is a device that detects a machine machining condition in the cutting wire section 6a. The machine machining condition detection device 15 is installed on wires 19 that connect the energy supply element 12 and the machine machining energy source 7, and detects a machine machining condition in the cutting wire section 6a by monitoring, via the energy supply element 12, the state of the voltage (hereinafter referred to as an electrode gap voltage) that is applied to the electrode gap between the workpiece 2 and the cutting wire section 6a, the number of spark erosion events per unit of time, the number of short circuits per unit of time, and a machine machining pulse voltage, which is an output voltage of the machine machining energy source 7.Furthermore, the machine processing condition detection device 15 comprises several machine processing condition detection units 16, such that there is a one-to-one correspondence between the machine processing condition detection units 16 and the cutting wires of the cutting wire area 6a, and detects the machine processing condition of the cutting wire area 6a for each cutting wire.

[0022] A connection target 20a, which is one of the connection targets of the wire break detection unit 20, is connected via an energy supply element 20d for wire break detection to the wire electrode 6 between the wire coil 4 and the energy supply element unit 13, with which the wire electrode 6 supplied by the wire coil 4 is initially in sliding contact. Furthermore, a connection target 20b, which is the other of the connection targets of the wire break detection unit 20, is in sliding contact with the wire ejection roller 5. This means that the wire break detection unit 20 is electrically connected in parallel to a wire electrode section that extends over a region from a location on the feed side of the wire electrode 6 to a location on the ejection side of the wire electrode 6.In this case, the wire ejection roller 5 is made of a conductive material, and the connection target 20b of the wire break detection unit 20 and the ejection side of the wire electrode 6 are electrically connected via the wire ejection roller 5.

[0023] Next, a configuration of the wire break detection unit 20 and a principle of a method for detecting a wire break are described. Fig. Figure 2 is a diagram which, in the form of an equivalent circuit, shows the relationship between the wire break detection unit 20 and a wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20 during multi-wire electrical discharge machining (EDM). During multi-wire EDM machining, spark erosion occurs stochastically in the electrode gap between the workpiece 2 and the cutting wire section 6a, depending on the state of the electrode gap (for example, the specific resistance of the machining fluid, the local temperatures of the workpiece 2 and the cutting wire, and the amplitude of the machining pulse voltage). This means that the voltage generated in the wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20 (i.e.,, a signal input to the wire break detection unit 20) can be considered as a simulated voltage source 60, which fluctuates randomly according to the machine machining pulse voltage and the state of an EDM operation. This means that a high-frequency voltage, which fluctuates randomly according to the machine machining pulse voltage and the state of the EDM operation, is generated in the wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20. Therefore, the relationship between the wire break detection unit 20 and the wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20 during multi-wire EDM machine machining can be described by the equivalent circuit in . Fig. 2 will be shown.

[0024] Fig. Figure 3 is a circuit diagram showing a configuration of the wire break detection unit 20 of the present embodiment. The wire break detection unit 20 comprises a wire break detection circuit 206, which is an electrical circuit in which a wire break detection energy source 204 and a current-limiting resistor 205 are connected in series. The wire break detection energy source 204 supplies a direct current to the wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20.

[0025] At least during the execution of a cutting process on workpiece 2, the wire break detection energy source 204 applies a constant DC voltage as a break detection voltage, which is lower than the machine processing pulse voltage to be applied by the machine processing energy source 7, and supplies a DC current to the wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20. This means that a constant DC current is supplied to the wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20 at least during the execution of a cutting process on workpiece 2. The DC current for wire break detection is approximately a few mA to a few hundred mA, which is sufficiently smaller than a machine processing current of approximately a few A to a few hundred A, and therefore does not affect the machine processing.The current-limiting resistor 205 limits the current flowing through the wire break detection circuit 206. This means that the current-limiting resistor 205 limits the direct current supplied by the wire break detection energy source 204 to the wire electrode section between connection target 20a and connection target 20b of the wire break detection unit 20, and also limits the current supplied to the wire break detection unit 20 by the simulated voltage source 60. Furthermore, the wire break detection unit 20 includes a wire break determination unit 207, which determines a wire break based on the current flowing through the current-limiting resistor 205. The wire break determination unit 207 outputs a wire break signal, indicating the result of the determination, to the machine control unit 9 to communicate a wire break detection result to the machine control unit 9.

[0026] Next, a detailed description is given of how the wire break detection unit 20 detects a wire break. In the Fig. In the circuit shown in Figure 3, the current flowing through the current-limiting resistor 205 is a current resulting from the superposition of a direct current output by the wire break detection energy source 204 and a high-frequency current output by the simulated voltage source 60. The wire break detection energy source 204 acts as a direct voltage source; therefore, if the wire electrode 6 is not broken, the wire break detection energy source 204 causes a direct current to flow through the current-limiting resistor 205. Because the simulated voltage source 60 inputs a random voltage signal into the wire break detection unit 20, a high-frequency current also flows from the simulated voltage source 60 into the current-limiting resistor 205.

[0027] The current to be detected by the wire break detection unit 207 is therefore a current resulting from the superposition of the direct current output by the wire break detection energy source 204 with the high-frequency current output by the simulated voltage source 60. However, if the wire electrode 6 is broken, the connection targets 20a and 20b of the wire break detection unit 20 are not electrically connected. Therefore, no current flows through the current-limiting resistor 205. In a case where a state can be detected in which current has ceased to flow through the current-limiting resistor 205, the wire break detection unit 207 determines that a wire break has occurred. In this way, it is possible to detect a wire break by sequentially monitoring the current flowing through the current-limiting resistor 205.

[0028] It should be noted that instead of the current-limiting resistor 205, the wire electrode 6, which connects the connection targets 20a and 20b of the wire break detection unit 20, can be used with a sufficient length, and the circuit configuration of the current-limiting resistor 205 is not limited as long as it is possible to detect a wire break by sequentially monitoring the current flowing through the current-limiting resistor 205.

[0029] Next, a hardware configuration of the machine processing control unit 9 of the multi-wire electrical discharge machining (EDM) machine 1 according to the first embodiment is described. Each function of the machine processing control unit 9 is implemented using a computer system, such as a personal computer or a general-purpose computer. Fig. Figure 4 is a diagram showing an example of a hardware configuration of the machine processing control unit 9 in a case in which the functions of the machine processing control unit 9 according to the first embodiment are implemented using a computer system.

[0030] The machine processing control device 9 comprises a processor 91, a memory 92, an external storage device 93, an input unit 94, and an output unit 95. The processor 91 performs various processing operations. The memory 92 is built-in memory. The external storage device 93 stores various types of information. The input unit 94 receives information entered by an operator. The output unit 95 displays various types of information back to the operator.

[0031] The processor 91 is a central processing unit (CPU). The processor 91 can be a processing device, an arithmetic unit, a microprocessor, a microcomputer, or a digital signal processor (DSP). Each function of the machine processing control unit 9 is implemented by the processor 91 and software, firmware, or a combination of software and firmware. The software or firmware is defined as a program and stored in the external storage device 93. The processor 91 reads the software or firmware stored in the external storage device 93 into memory 92 and executes the software or firmware.

[0032] Memory 92 is a non-volatile or volatile semiconductor memory. Examples of memory 92 include random-access memory (RAM), read-only memory (ROM), flash memory, EPROM (Erasable Programmable Read Only Memory), and EEPROM (registered trademark) (Electrically Erasable Programmable Read Only Memory). External storage device 93 is a hard disk drive (HDD) or solid-state drive (SSD).

[0033] The input unit 94 comprises an input device represented by an input key and a pointing device. The input unit 94 receives information entered by an operator of the multi-wire electrical discharge machining (EDM) machine 1 and transmits this information to the processor 91. The output unit 95 comprises an output device represented by a liquid crystal display (LCD) and a loudspeaker. The output unit 95, together with the pointing device of the input unit 94, forms a single touchscreen. The output unit 95 displays various types of information to the operator according to instructions from the processor 91.

[0034] As described above, in such a configuration, the multi-wire electrical discharge machining (EDM) machine 1 according to the present embodiment applies the DC voltage output by the wire break detection energy source 204 to the wire electrode 6 at a constant rate, at least during a cutting process on the workpiece 2, and checks the conductivity of the wire electrode 6. This allows the multi-wire EDM machine 1 according to the present embodiment to accurately detect a wire break without being affected by inductance generated by the winding of the single wire electrode 6 around the multiple guide rollers.Furthermore, it is possible to quickly detect a wire break by applying the DC voltage output by the wire break detection energy source 204 to the wire electrode 6 and by checking the conductivity of the wire based on the current resulting from the superposition of the high-frequency current input from the wire electrode 6 into the wire break detection unit 20 with the DC current provided by the wire break detection energy source 204, not only during a period in which no machine processing pulse voltage is applied by the machine processing energy source 7, but also during multi-wire electrical discharge machining. Second embodiment.

[0035] In the first embodiment, the Fig. The configuration of the wire break detection unit 20 shown in Figure 3 is described. However, the configuration of the wire break detection unit 20 can be as shown in Figure 3. Fig. 5 will be shown. The configuration of Fig. 5 is the same as the configuration of Fig. 3 with the difference that the configuration of Fig. Three rectifier circuits 201 and 202 and a smoothing circuit 203 were added. The rectifier circuits 201 and 202 rectify a high-frequency voltage generated in the wire electrode section between the connection target 20a and the connection target 20b of the wire break detection unit 20, according to the machine processing pulse voltage and the condition of the electrical discharge machining (EDM). The smoothing circuit 203 smooths a current rectified by the rectifier circuits 201 and 202. The rectifier circuits 201 and 202 only need to be capable of rectifying a high-frequency voltage and include, for example, rectifier diodes 201a and 202a. The smoothing circuit 203 only needs to be capable of smoothing the current rectified by the rectifier circuits 201 and 202, and includes, for example, a circuit in which a high-frequency cut-off inductor 203a and a damping diode 203b are connected in parallel.With this configuration, a high-frequency current output by the simulated voltage source 60 is smoothed and superimposed on a direct current output by the wire break detection energy source 204, and the resulting current flows through the current-limiting resistor 205. The wire break detection unit 207 detects this current and checks the conductivity of the wire using the same method as in the first embodiment. This allows a wire break to be detected. Third embodiment.

[0036] In a case where the conductivity of the wire electrode 6 is tested using a current flowing through the current-limiting resistor 205, the current flowing through the current-limiting resistor 205 oscillates immediately after a wire break occurs. To accurately detect a wire break, it was therefore necessary to wait until the current had stabilized before detection. In a third embodiment, a method for rapidly detecting a wire break is described. Parts that are the same as those of the first embodiment are not described, and parts that differ from those of the first embodiment are described.

[0037] Fig. Figure 6 is a circuit diagram showing a configuration of a wire break detection unit 30 of a multi-wire electrical discharge machining (EDM) machine according to the third embodiment. A wire break detection unit 305 comprises a binary circuit 304 and a signal processing circuit 303. The binary circuit 304 converts a current flowing through the current-limiting resistor 205 into a binary signal, which is either a low-level signal or a high-level signal, depending on whether current flows through the current-limiting resistor 205. The signal processing circuit 303 outputs a wire break signal based on the binary signal. For example, it is possible to configure the binary circuit 304 as a combination of a photocoupler 304a and a pull-up resistor 304b, as shown in Figure 6. Fig. Figure 6 shows that the binaryization circuit 304 converts a current flowing through the current-limiting resistor 205 into a binary signal by outputting a low-level analog signal when current flows through the current-limiting resistor 205, and outputting a high-level analog signal when no current flows through the current-limiting resistor 205.

[0038] The signal processing circuit 303 outputs a wire break signal based on the result of a detection of whether the signal output by the binarization circuit 304 is a high-level or low-level signal. In this way, it is possible to detect a wire break by monitoring the wire break signal. In the multi-wire electrical discharge machine according to the third embodiment, the signal processing circuit 303 comprises a low-pass filter unit 301 and a wire break signal output unit 302 to detect whether the analog signal output by the binarization circuit 304 is a high-level or low-level signal. The low-pass filter unit 301 is a circuit with a low-pass characteristic. The wire break signal output unit 302 outputs a wire break signal based on a signal output by the low-pass filter unit 301.Examples of configurations for the Low-Pass Characteristic Unit 301 include a CR low-pass filter, an active low-pass filter using an operational amplifier, a digital filter using a D flip-flop, and a finite impulse response (FIR) filter. However, the circuit configuration of the Low-Pass Characteristic Unit 301 is not limited to such configurations, as long as the Low-Pass Characteristic Unit 301 is a circuit with a low-pass characteristic.

[0039] Fig. Figure 7 is a diagram describing the response of the binary circuit 304 before and after a wire break. Before the wire break (between time "a" and time "b"), the connection targets 20a and 20b of the wire break detection unit 30 are brought into a conducting state by the wire electrode 6, and current flows through the current-limiting resistor 205, so the output of the binary circuit 304 indicates a low level. Although the output current of the wire break detection power supply 204 becomes zero immediately after the wire break occurs (between time "b" and time "c"), the output current of the simulated voltage source 60 temporarily flows into the wire break detection unit 30 before converging to zero, causing a binary signal to oscillate and reach the high level. Fig. Figure 8 is a diagram describing the response of the low-pass characteristic unit 301 before and after the wire break. When an oscillating binary signal is input into the low-pass characteristic unit 301, an oscillating component is removed by a low-pass filter, resulting in a signal waveform as shown in Fig. Figure 8 shows that when a signal passing through the low-pass characteristic unit 301 reaches or exceeds a preset threshold, the wire break signal output unit 302 determines that a wire break has occurred and outputs a wire break signal. This means that the signal processing circuit 303 outputs a wire break signal based on the result of a comparison between the predetermined threshold and a signal, which is a binary signal that has passed through the low-pass filter.

[0040] As described above, according to the present embodiment, the wire break detection unit 305 comprises the binary conversion circuit 304 and the signal processing circuit 303. The binary conversion circuit 304 converts a current flowing through the current-limiting resistor 205 into a binary signal, which is either a low-level or a high-level signal, depending on whether a current is flowing through the current-limiting resistor 205. The signal processing circuit 303 has a low-pass characteristic. This makes it possible to quickly detect a wire break even if the current flowing through the current-limiting resistor 205, which is being monitored to check a line condition, oscillates immediately after the wire break occurs. Fourth embodiment.

[0041] In the third embodiment, a method for rapidly detecting a wire break is described, even in the case where a current flowing through the current-limiting resistor 205 oscillates immediately after the wire break occurs. This method utilizes the wire break detection unit 305, which is configured to include the binaryization circuit 304 and the low-pass signal processing circuit 303. A fourth embodiment describes a method for rapidly detecting a wire break even with a configuration in which the signal processing circuit does not have a low-pass characteristic. Parts identical to those of the third embodiment are not described, while parts differing from those of the third embodiment are described.

[0042] Fig. Figure 9 is a circuit diagram showing the configuration of a wire break detection unit 40 of a multi-wire electrical discharge machining (EDM) machine according to the fourth embodiment. The multi-wire EDM machine according to the fourth embodiment has the same configuration as the multi-wire EDM machine according to the third embodiment, except that the multi-wire EDM machine according to the fourth embodiment includes a signal processing circuit 403, which differs from the signal processing circuit of the third embodiment. The signal processing circuit 403 includes a frequency detection unit 401 and a wire break signal output unit 402. The frequency detection unit 401 detects the frequency of a signal output by the binaryization circuit 304. The wire break signal output unit 402 outputs a wire break signal based on the detected frequency.As described above, a binary signal immediately after a wire break occurs has a behavior similar to that shown in . Fig. 7. Waveform shown, in which an oscillation repeats periodically. Fig. Figure 10 is a diagram describing the response of the frequency detection unit 401 before and after a wire break. When a binary signal with a repeating periodic oscillation is input to the frequency detection unit 401, the unit detects the frequency of the binary signal and outputs a result as shown in Figure 10. Fig. The detection result shown in Figure 10 is sent to the wire break signal output unit 402. If the detected frequency is higher than a predetermined threshold (for example, several tens of Hz to several kHz), the wire break signal output unit 402 determines that a wire break has occurred and outputs a wire break signal. This means that the signal processing circuit 403 detects the frequency of a binary signal and outputs a wire break signal based on a comparison with the predetermined threshold.

[0043] As described above, according to the present embodiment, a wire break is determined based on the detection of the frequency of a binary signal, so that it is possible to quickly detect a wire break even if a current flowing through the current-limiting resistor 205, which is monitored to check a line condition, oscillates immediately after the wire break occurs. Fifth embodiment.

[0044] In the first through fourth embodiments, a method for detecting a wire break upon its occurrence was described. However, in a fifth embodiment, a method is described that predicts a wire break in advance and prevents it using the results of machine learning. Parts that are identical to those of the first through fourth embodiments are not described, while parts that differ are described.

[0045] Fig. Figure 11 is a perspective view showing a configuration of a main part of a multi-wire electrical discharge machining (EDM) machine 1a according to the fifth embodiment. The multi-wire EDM machine 1a is configured such that a wire break prediction unit 21 and a machine learning device 22 are added to the configuration of the main part of the multi-wire EDM machine 1 of the first embodiment.

[0046] Next, the operation of the wire break prediction unit 21 and the machine learning device 22 is described separately for a learning phase and an application phase. Learning phase

[0047] Fig. Figure 12 is a diagram showing a configuration of the machine learning device 22 according to the fifth embodiment. The machine learning device 22 comprises a data reference unit 221, a state observation unit 222, a model generation unit 223, and a learned model storage unit 224.

[0048] The data reference unit 221 receives a wire break detection result from the machine processing control unit 9, which is output by the wire break detection unit 20. The condition monitoring unit 222 receives condition data from the machine processing control unit 9 as information indicating a machine processing state during machining. Examples of the condition data include the relative distance between the workpiece 2 and the cutting wire area 6a, a wire movement speed, a wire tension, and several results, which are output by the machine processing state detection device 15.

[0049] The model generation unit 223 learns, based on training data generated from combinations of wire break detection results output by the data reference unit 221 and state data output by the state monitoring unit 222, information (hereinafter referred to as prediction information) indicating the probability that the wire electrode 6 will break. This means that a learned model is generated which derives optimal prediction information from the wire break detection results and the state data. In this case, the training data consists of data in which wire break detection results and state data are associated with each other.

[0050] It should be noted that the machine learning device 22 is used to learn the predictive information, but it can, for example, be a device that is provided separately from the multi-wire electrical discharge machine and connected to the multi-wire electrical discharge machine via a network. Furthermore, the machine learning device 22 can be integrated into the multi-wire electrical discharge machine. Finally, the machine learning device 22 can reside on a cloud server. Fig. Figure 16 is a diagram showing a configuration of a multi-wire electrical discharge machining (EDM) system 500 with a multi-wire EDM machine 1b and a machine learning device 22b. The present multi-wire EDM system 500 has a configuration that differs from the configuration in Fig. 11 differs in that the machine learning device 22b is located outside the multi-wire electrical discharge machine 1b. The machine learning device 22b receives wire break detection results and condition data from the multi-wire electrical discharge machine 1b as training data and generates a learned model through machine learning. The wire break prediction unit 21 of the multi-wire electrical discharge machine 1b receives the learned model from the machine learning device 22b and outputs a prediction using the learned model.

[0051] Referring again to Fig. 12. A known supervised learning algorithm can be used by the model generation unit 223 as a learning algorithm. As an example, a case is described in which a neural network is used. For instance, the model generation unit 223 learns predictive information through so-called supervised learning with a neural network. In this context, supervised learning refers to a technique of providing datasets consisting of inputs and outputs (labels) to a learning device in order to learn features in this training data and derive an output from an input.

[0052] The neural network comprises an input layer with multiple neurons, an intermediate layer (hidden layer) with multiple neurons, and an output layer with multiple neurons. The neural network can have a single intermediate layer or it can have two or more intermediate layers.

[0053] In the case of a three-layer neural network, as in Fig. As shown in Figure 13, for example, if multiple inputs are provided at an input layer (X1 to X3), the values ​​are multiplied by weights W1 (w11 to w16) and fed into an intermediate layer (Y1 to Y2), and the results are further multiplied by weights W2 (w21 to w26) and output by an output layer (Z1 to Z3). The output results vary depending on the values ​​of weights W1 and W2.

[0054] In the present application, the neural network learns predictive information using so-called supervised learning based on training data, which is generated based on combinations of wire break detection results obtained by the data reference unit 221 and state data obtained by the state observation unit 222.

[0055] This means that the neural network performs learning by adjusting the weights W1 and W2 in such a way that a result output from the output layer in response to the input of state data into the input layer approximates a wire break detection result.

[0056] The model generation unit 223 generates a learned model by performing the learning process described above and outputs it.

[0057] The learned model storage unit 224 stores the learned model output by the model generation unit 223.

[0058] Next, with reference to Fig. 14 describes a learning process which is to be carried out by the machine learning device 22. Fig. Figure 14 is a flowchart concerning a learning process to be carried out by the machine learning device 22.

[0059] In step b1, the data reference unit 221 receives a wire break detection result, and the condition monitoring unit 222 receives condition data. Although the wire break detection result and the condition data are assumed to be obtained simultaneously, it should be noted that the wire break detection result and the condition data can be obtained at different times, as long as the wire break detection result and the condition data can be entered in association with each other.

[0060] In step b2, the model generation unit 223 creates a learned model by learning predictive information through so-called supervised learning based on training data, which are generated based on combinations of wire break detection results obtained by the data reference unit 221 and the state data obtained by the state observation unit 222.

[0061] In step b3, the learned model storage unit 224 stores the learned model generated by the model generation unit 223. Application phase

[0062] Fig. Figure 15 is a configuration diagram of the wire break prediction unit 21 relating to the multi-wire electrical discharge machining (EDM) machine. The wire break prediction unit 21 comprises a condition monitoring unit 211 and a derivation unit 212.

[0063] The condition monitoring unit 211 receives condition data.

[0064] The derivation unit 212 derives predictive information obtained using a learned model. This means that, as a result of inputting the state data obtained by the state observation unit 211 into the learned model, it is possible to output predictive information derived from the state data.

[0065] It should be noted that in the present embodiment it has been described that the prediction information is output using a learned model, which is obtained as a result of the learning carried out by the model generation unit 223 of the multi-wire electrical discharge machining (EDM) machine 1a; however, a learned model can be obtained from an external device, such as another multi-wire EDM machine, and prediction information can be output based on the learned model.

[0066] Next, processing to obtain predictive information using the wire break prediction unit 21 with reference to Fig. 18 described.

[0067] In step c1, the state monitoring unit 211 obtains state data.

[0068] In step c2, the wire break prediction unit 21 inputs the state data into a learned model, which is stored in the learned model storage unit 224, in order to obtain prediction information.

[0069] In step c3, the derivation unit 212 outputs the prediction information obtained using the learned model to the machine processing control unit 9.

[0070] In step c4, the machining control unit 9 controls a wire EDM machining operation using the output prediction information to prevent wire electrode breakage. If the prediction information indicates a possibility of wire breakage, the machining control unit 9 instructs the machining power source 7 to reduce the machining power. This means that the machining control unit 9 controls the machining power source 7 so that the machining power becomes lower than the current machining power. Furthermore, the machining control unit 9 can issue a command to decrease the wire tension or a command to increase the wire feed speed to the guide roller drive unit 18 and the wire spool drive unit 17.This means that the machine control unit 9 controls the guide roller drive unit and the wire spool drive unit so that the wire tension becomes lower than the current wire tension or the wire feed speed becomes higher than the current wire feed speed. Therefore, wire breakage during multi-wire EDM machining can be prevented, and the workpiece 2 can be reliably machined.

[0071] It should be noted that the present embodiment describes the case in which supervised learning is used as a learning algorithm by the model generation unit 223; however, the learning algorithm is not limited to this. Semi-supervised learning or the like can also be used as the learning algorithm in addition to supervised learning.

[0072] Furthermore, the Model Generation Unit 223 can learn predictive information based on training data generated for multiple multi-wire EDM machines. It should be noted that the Model Generation Unit 223 can obtain training data from multiple multi-wire EDM machines operating in the same area, or it can learn predictive information using training data collected from multiple multi-wire EDM machines operating independently in different areas. Additionally, a multi-wire EDM machine from which training data is collected can be added to or removed from the group of multi-wire EDM machines during the learning process.Furthermore, a machine learning device that has learned predictive information for a particular multi-wire electrical discharge machining (EDM) machine can be used on another multi-wire EDM machine, and the predictive information can be relearned and updated for the other multi-wire EDM machine.

[0073] Furthermore, deep learning, in which the extraction of a feature set is learned, can be used by the model generation unit 223 as the learning algorithm. Alternatively, machine learning can be performed according to another known method, such as genetic programming, functional logic programming, or a support vector machine.

[0074] In the example described above, prediction information is calculated within the multi-wire EDM machines 1a and 1b, however, the prediction information can be generated by a wire break prediction device which is different from the multi-wire EDM machines 1a and 1b. Fig.Figure 17 is a diagram showing a configuration of a multi-wire electrical discharge machining (EDM) processing system 600 with a multi-wire EDM machine 1c and a wire break prediction device 23. The wire break prediction device 23 comprises the wire break prediction unit 21 and the machine learning device 22, which are identical to those described above. During the learning phase, the wire break prediction device 23 receives wire break detection results and condition data from the multi-wire EDM machine 1c as training data and generates a learned model by performing processing identical to that described in the "Learning Phase" section above.During the application phase, the wire break prediction device 23 calculates prediction information by reacquiring state data from the multi-wire electrical discharge machining (EDM) machine 1c and inputting this data into the learned model generated during the learning phase. The predictive information is then output to the machining control unit 9 of the multi-wire EDM machine 1c. The machining control unit 9 controls the EDM machining operation using a method identical to the one described above, thus preventing wire electrode breakage using the acquired prediction information. With this configuration of the multi-wire EDM machining system 600, the machine learning and derivation processing, which would otherwise burden the system, can be performed outside the multi-wire EDM machine.

[0075] According to the fifth embodiment, it is possible to achieve the effect of preventing wire breakage by controlling machine processing conditions based on a prediction of wire breakage during multi-wire electrical discharge machining (EDM). Furthermore, the machine learning phase can be executed for each wire breakage. This improves the accuracy of wire breakage prediction.

[0076] The configurations set forth in the above embodiments show examples of the subject matter of the present disclosure, and it is possible to combine the configurations with another known technique, and it is also possible to partially omit or modify the configurations without departing from the scope of the present disclosure. Reference symbol list 1, 1a, 1b, 1c Multi-wire electrical discharge machine; 2 workpieces; 3a, 3b, 3c, 3d Leadership role; 4 wire coils; 5 wire ejection roller; 6 wire electrode; 6a Cutting wire area; 6b Power supply wire area; 7. Machine processing energy source; 8 Machine processing power supply unit; 9 Machine processing control unit; 10 Ground electrode; 11 cables; 12 Energy supply element; 13 Energy supply element unit; 14 Drive unit; 15 Machine processing condition detection device; 16 Machine processing condition detection unit; 17 Wire spool drive unit; 18 Guide roller drive unit; 19 wire; 20, 30, 40, 50 wire break detection unit; 20a, 20b Connection target of the wire break detection unit; 20d Energy supply element for wire break detection; 21 Wire break prediction unit; 22, 22b Machine learning device; 23 Wire break prediction device; 60 simulated voltage source; 70 machine work table; 91 processor; 92 memory slots; 93 external storage device; 94 Input unit; 95 output units; 201,202 Rectifier circuit; 201a, 202a Rectifier diode; 203 Smoothing circuit; 203a High-frequency shutdown inductor; 203b Damping diode; 204 Wire break detection energy source; 205 current-limiting resistor; 206 Wire break detection circuit; 207, 305, 404 Wire breakage determination unit; 211,222 Condition monitoring unit; 212 derivation unit; 221 Data reference unit; 223 Model generation unit; 224 Learned Model Memory Unit; 301 Low-pass characteristic unit; 302,402 Wire break signal output unit; 303,403 Signal processing circuit; 304 Binary circuit; 304a Photocoupler; 304b Pull-Up Resistor; 401 Frequency detection unit; 500, 600 multi-wire electrical discharge machining (EDM) machining system.

Claims

[1] Multi-wire electrical discharge machining (EDM) machine comprising a cutting wire section which includes several cutting wires arranged in parallel and facing a workpiece, the several cutting wires extending parallel between several guide rollers by having a single wire electrode wound around the several guide rollers, and which cuts the workpiece by causing electrical discharge between the cutting wire section and the workpiece, the multi-wire EDM machine comprising: a machining energy source that applies a machining pulse voltage between the workpiece and the cutting wire area; and a wire break detection unit which detects a break in the wire electrode, wherein the wire break detection unit is electrically connected in parallel to a wire electrode section which extends over an area from a location on a feed side of the wire electrode to a location on an ejection side of the wire electrode, and The wire break detection unit includes: a wire break detection circuit, which is an electrical circuit in which a wire break detection energy source and a current-limiting resistor are connected in series, wherein the wire break detection energy source supplies a constant direct current to the wire electrode section at least during one execution of a cutting process on the workpiece; and a wire break detection unit which determines a break in the wire electrode based on a current flowing through the current-limiting resistor. [2] Multi-wire electrical discharge machining (EDM) machine according to claim 1, wherein the wire break detection circuit is an electrical circuit in which a rectifier circuit and a smoothing circuit are connected in series, wherein the rectifier circuit rectifies a high-frequency voltage generated in the wire electrode section according to the machine processing pulse voltage and a state of EDM, wherein the smoothing circuit smooths a current rectified by the rectifier circuit. [3] Multi-wire electrical discharge machine according to claim 1 or 2, wherein, when a condition in which a current ceases to flow through the current-limiting resistor can be detected, the wire break detection unit determines that the wire electrode is broken. [4] Multi-wire electrical discharge machine according to claim 3, wherein the wire break detection unit comprises: a binary conversion circuit which converts a current flowing through the current-limiting resistor into a binary signal which, depending on whether current flows through the current-limiting resistor, is a low-level signal or a high-level signal; and A signal processing circuit which, based on the binary signal, detects a state in which current ceases to flow through the current-limiting resistor and outputs a wire break signal, which indicates a result of the determination of a break in the wire electrode. [5] Multi-wire electrical discharge machine according to claim 4, wherein the signal processing circuit outputs the wire break signal based on the result of a comparison between a predetermined threshold and a signal obtained by passing the binary signal through a low-pass filter. [6] Multi-wire electrical discharge machine according to claim 4, wherein the signal processing circuit detects a frequency of the binary signal and outputs the wire break signal based on a result of a comparison with a predetermined threshold. [7] Multi-wire electrical discharge machining machine according to any one of claims 1 to 6, further comprising: a wire break prediction unit which outputs predictive information which is information indicating a possibility of a wire electrode break, wherein The wire break prediction unit includes: a condition monitoring unit which receives condition data, wherein the condition data is information indicating a machine processing state; and a derivation unit which derives the prediction information from the state data input by the state observation unit using a learned model generated by machine learning based on combinations of the state data and results of wire electrode break detection. [8] Multi-wire electrical discharge machine according to claim 7, wherein the state data comprise at least one of the following: a relative distance between the workpiece and the cutting wire area, a wire movement speed, a wire tension an electrode gap voltage, the number of spark erosion events per unit of time, the number of short circuits per unit of time and the machine processing pulse voltage. [9] Multi-wire electrical discharge machine according to claim 7 or 8, further comprising: a machine learning device that learns the predictive information, wherein the machine learning device comprises: a condition monitoring unit that receives the condition data; a data reference unit which receives a result of the detection of the wire electrode breakage; and a model generation unit that generates the learned model through machine learning based on training data, which are generated based on combinations of the state data and the results of the detection of the wire electrode breakage. [10] Multi-wire electrical discharge machine according to any one of claims 7 to 9, further comprising: a machine processing control unit which controls a spark erosion machine processing operation based on the prediction information in such a way as to prevent breakage of the wire electrode. [11] Multi-wire electrical discharge machining machine according to claim 10, wherein, when the prediction information indicates that there is a possibility of wire electrode breakage, the machine processing control unit controls the machine processing energy source such that a machine processing energy becomes less than a current machine processing energy. [12] Multi-wire electrical discharge machining machine according to claim 10 or 11, further comprising: a guide roller drive unit which drives the guide roller; and a wire spool drive unit which drives a wire spool, wherein, If the prediction information indicates that there is a possibility of wire electrode breakage, the machine processing control unit controls the guide roller drive unit and the wire spool drive unit so that a wire tension becomes less than a current wire tension or that a wire movement speed becomes greater than a current wire movement speed. [13] Multi-wire electrical discharge machining system comprising: the multi-wire electrical discharge machine according to claim 7 or 8; and a machine learning device that generates the learned model, whereby The machine learning tool includes: a condition monitoring unit that receives the condition data; a data reference unit which receives a result of the detection of the wire electrode breakage; and a model generation unit that generates the learned model through machine learning based on training data, which are generated based on combinations of the state data and the results of the detection of the wire electrode breakage. [14] Multi-wire electrical discharge machining system comprising: the multi-wire electrical discharge machining (EDM) machine according to any one of claims 1 to 6; and a wire break prediction device which outputs predictive information which is information indicating a possibility of a wire electrode break, wherein The wire break prediction device includes: a condition monitoring unit that receives condition data, where the status data is information that indicates a machine processing state; a data reference unit which receives a result of the detection of the breakage of the wire electrode; a model generation unit that creates a learned model through machine learning based on training data, which are generated based on combinations of the state data and the results of the detection of the wire electrode breakage; and a derivation unit that derives the prediction information from the state data using the learned model, where the condition data is entered by the condition monitoring unit. [15] Multi-wire electrical discharge machining system according to claim 14, wherein the state data comprise at least one of the following: a relative distance between the workpiece and the cutting wire area, a wire movement speed, a wire tension an electrode gap voltage, the number of spark erosion events per unit of time, the number of short circuits per unit of time and the machine processing pulse voltage.

Images