Die maintenance system, die quality determination threshold value generation device, learning device, die quality determination method, and program

Abstract

A die maintenance system (100) comprises: a TPP processing machine (1) that is provided with a plurality of dies and that selectively uses a die according to an NC program to perform processing; an NC data monitoring device (2) that acquires NC data which indicates the content of processing of the TPP processing machine (1); a sensing device (3) that generates measurement data obtained by measuring processing sounds which are generated when the die is used by the TPP processing machine (1); a die quality determination threshold value generation device (5) that, on the basis of the measurement data, generates threshold value data which serves as a reference for determining the quality of the die of the TPP processing machine (1); and a die quality determination device (6) that, on the basis of the NC data, the measurement data, and the threshold value data which serves as the reference for determining the quality of the die, determines the quality of the die of the TPP processing machine (1).

Description

Mold maintenance system, mold quality determination threshold value generation device, learning device, mold quality determination method and program

[0001] The present disclosure relates to a mold maintenance system, a mold quality determination threshold value generation device, a learning device, a mold quality determination method, and a program.

[0002] At sheet metal processing sites, there are automatic processing machines called TPP (Turret Punch Press) machines, which are equipped with hundreds of general-purpose dies and can punch out specified shapes. Because TPP machines have a large number of general-purpose dies, they are used for high-mix, low-volume production, such as individual order production. With TPP machines, the condition of wear on the edges of the dies (die and punch) has a significant impact on the quality of the processed product and the productivity of the equipment, so die inspections are carried out.

[0003] When inspecting the dies of a TPP processing machine, the state of die wear is checked, and if it is determined that wear is severe, maintenance or replacement is performed. The number of times the die has been used is generally used as an indicator of die wear. Hereinafter, the number of times the die has been used is referred to as the punch count. The punch count can be obtained from the equipment. However, TPP processing machines use a variety of die-based processing techniques, such as overtaking and nibbling. For example, with overtaking, only a portion of the die is used, limiting the area of ​​wear. Furthermore, the frequency with which each processing method is applied varies depending on the product being processed. For this reason, the punch count alone does not accurately determine the state of die wear or the appropriate inspection time.

[0004] There are two main problems with die inspections: over-inspection and under-inspection. Over-inspection is performing maintenance on dies that are still usable. Ideally, they could continue to be used in production without replacement, but since they become unusable, production using that equipment cannot be carried out during replacement, resulting in reduced productivity. Under-inspection is when wear that cannot be determined by punch count alone progresses, and necessary maintenance is not carried out. As die wear progresses, burrs are more likely to occur on the cut surface of the workpiece, resulting in significant losses due to poor quality. Furthermore, processing with excessively worn dies can cause equipment shutdowns due to die damage.

[0005] Therefore, currently, an experienced person must remove the die from the equipment and directly inspect the processed surface to determine wear. However, die inspection by an experienced person requires equipment shutdowns and worker man-hours each time an inspection is performed, increasing costs. It is also necessary to secure craftsmen who can properly inspect dies in the sheet metal workplace. Furthermore, determining die wear is subjective, and failure to properly inspect dies can lead to equipment shutdowns due to reduced quality and poor processing. For this reason, there is a demand for technology that can detect die wear and automatically determine whether a die is good or bad.

[0006] Patent Document 1 discloses a technique for detecting die wear in a TPP processing machine from characteristic data of punching sounds.

[0007] Japanese Patent Application Publication No. 4-42024

[0008] In a TPP processing machine, the processing conditions (mold shape, processing method, plate thickness, material, etc.) for the mold and workpiece are different each time. Therefore, if the technology described in Patent Document 1 is applied to mold inspection in a TPP processing machine, a standard for determining whether the mold is good or bad is required for each processing condition. For example, when processing sheet metal with five different plate thicknesses, five different materials, and 300 different molds, it is necessary to prepare 7,500 patterns (5 x 5 x 300) of mold quality determination standards. However, this requires a large amount of cost for preparation and data acquisition to set the mold quality determination standards, making it impractical.

[0009] The present disclosure has been made to solve the problems described above, and aims to enable automatic determination of the quality of dies in TPP processing machines and reduce the cost of setting standards for determining the quality of dies.

[0010] To achieve the above object, the mold maintenance system according to the present disclosure includes a TPP processing machine, an NC data monitoring device, a sensing device, a mold quality determination standard generation device, and a mold quality judgment device. The TPP processing machine is equipped with multiple molds and performs processing by selectively using the molds according to an NC program. The NC data monitoring device acquires NC data indicating the processing content of the TPP processing machine. The sensing device measures processing sounds generated when the molds are used in the TPP processing machine and generates measurement data. The mold quality determination standard generation device generates threshold data or a trained model based on the measurement data, which serves as a standard for determining the quality of the molds in the TPP processing machine. The mold quality judgment device judges the quality of the molds in the TPP processing machine based on the NC data, the measurement data, and the threshold data or the trained model.

[0011] According to the present disclosure, threshold data or a trained model that serves as the standard for determining whether a die of a TPP processing machine is good or bad is automatically generated based on measurement data, and a determination of whether the die of a TPP processing machine is good or bad is made based on NC data and measurement data and the threshold data or trained model, thereby making it possible to automatically determine whether a die of a TPP processing machine is good or bad, and reducing the cost of setting standards for determining whether a die is good or bad.

[0012] FIG. 1 shows an example of the configuration of a mold maintenance system according to embodiment 1. FIG. 1 shows an example of the functional configuration of a TPP processing machine, an NC data monitoring device, and a sensing device according to embodiment 1. FIG. 2 shows an example of the execution processing data according to embodiment 1. FIG. 3 shows an example of mold quality determination reference data. FIG. 4 shows an example of the functional configuration of a processing condition setting device, a mold quality determination threshold generation device, and a mold quality determination device according to embodiment 1. FIG. 5 shows an example of the processing condition data according to embodiment 1. FIG. 6 shows a graph comparing the integration results of FFT analysis of sound waveforms with respect to the maximum value of the sound waveform of experimental data according to embodiment 1. FIG. 7 shows a graph comparing experimental data with different plate thicknesses according to embodiment 1. FIG. 8 shows a graph in which the first axis of the graph comparing experimental data with different plate thicknesses according to embodiment 1 is plate thickness. FIG. 1 shows an example of the configuration of a mold maintenance system according to embodiment 2. FIG. 2 shows an example of the functional configuration of an NC program change device according to embodiment 2. FIG. 3 shows an example of a mold replacement master according to embodiment 2. FIG. 4 shows an example of the operation of NC program change processing according to embodiment 2. FIG. 5 shows an example of the configuration of a mold maintenance system according to embodiment 3. FIG. 6 shows an example of the functional configuration of a learning device and a mold quality determination device according to embodiment 3. FIG. 7 shows the results of clustering experimental data according to embodiment 1 using the K-means method.

[0013] The die maintenance system, die quality determination threshold value generating device, learning device, die quality determination method, and program according to the first embodiment will be described in detail below with reference to the drawings. Note that identical or corresponding parts in the drawings are designated by the same reference numerals. In this embodiment, a "failed" die quality determination refers to a state in which a punching defect occurs due to wear on the edges of the die and punch, and replacement or polishing is required, regardless of whether the die is usable or not. Generally, as die wear progresses, shear stress increases, resulting in increased processing noise and impact during punching. The die maintenance system determines whether the die is good or bad by sensing the above-mentioned phenomena associated with die wear in a TPP processing machine.

[0014] (Embodiment 1) The configuration of a mold maintenance system 100 according to embodiment 1 will be described with reference to Fig. 1. The mold maintenance system 100 includes a TPP processing machine 1 that is equipped with multiple molds and selectively uses the molds according to an NC program to perform processing such as drilling and forming, an NC data monitoring device 2 that reads and stores NC data indicating the processing content displayed on the TPP processing machine 1, a sensing device 3 that measures processing sounds generated when the molds are used in the TPP processing machine 1, a processing condition setting device 4 that accepts input of processing conditions for the TPP processing machine 1, a mold quality determination threshold generation device 5 that generates threshold data indicating thresholds that serve as standards for determining whether the molds in the TPP processing machine 1 are good or bad, and a mold quality determination device 6 that uses the thresholds to determine whether the molds in the TPP processing machine 1 are good or bad. The NC data includes information such as NC codes indicating the processing content, the thickness of the workpiece, the material of the workpiece, and the processing date and time.

[0015] The functional configurations of the TPP machine 1, NC data monitoring device 2, and sensing device 3 will be described with reference to FIG. 2. The TPP machine 1 includes a press working unit 11 that performs processes such as drilling and forming according to an NC program, and an NC data display unit 12 that displays the NC data of the processing being performed by the press working unit 11. The press working unit 11 is a sheet metal processing device that positions the workpiece at the processing date and time included in the NC data, selectively uses dies based on the NC code, and performs processing such as drilling and forming. The NC data display unit 12 displays the NC data of the processing being performed by the press working unit 11. Hereinafter, the processing being performed by the press working unit 11 indicated by the NC data will be referred to as the "executed processing."

[0016] The NC data monitoring device 2 includes an NC data acquisition unit 21 that reads NC data, an NC data storage unit 22 that stores the NC data read by the NC data acquisition unit 21, a mold NC code master storage unit 23 that stores a mold NC code master that links NC codes with molds used in processing, and an execution processing data generation unit 24 that generates execution processing data by adding data on the molds used in processing to the NC data based on the NC data and the mold NC code master.

[0017] The NC data acquisition unit 21 is a camera that reads the NC data displayed on the NC data display unit 12 of the TPP machining machine 1. TPP machining machines are manufactured by a variety of manufacturers, and although the NC programs and NC data display units vary depending on the manufacturer, they all have the same basic configuration. Therefore, if the NC data acquisition unit 21 is a camera, it can read the displayed NC data regardless of the manufacturer of the TPP machining machine 1. The NC data acquisition unit 21 is not limited to a camera, and may also acquire NC data directly from the TPP machining machine 1. In this case, an interface must be developed for each manufacturer of the TPP machining machine 1. The NC data acquisition unit 21 stores the read NC data in the NC data storage unit 22.

[0018] The execution machining data generating unit 24 adds data on the dies used in machining to the NC data based on the NC data stored in the NC data storage unit 22 and the die NC code master stored in the die NC code master storage unit 23, and generates execution machining data including the type of die used in the execution machining, the machining method, the plate thickness of the workpiece, the material of the workpiece, the machining date and time, etc. The execution machining data generating unit 24 transmits the generated execution machining data to the die quality determination device 6.

[0019] Here, the execution machining data will be explained using FIG. 3. The example execution machining data shown in FIG. 3 includes the following fields: "NC Data," which indicates the NC data number of the processed sheet; "NC Code," which indicates the NC code included in the NC data read by the NC data acquisition unit 21; "Die Shape," which indicates the shape of the die used in the execution machining; "Machining Method," which indicates the machining method for the execution machining; "Material," which indicates the material of the workpiece for the execution machining; "Thickness," which indicates the thickness of the workpiece for the execution machining; and "Date and Time," which indicates the date and time of the execution machining. The "Die Shape" field is not limited to the die shape; any information that can identify the type of die is sufficient. For example, when the NC code is XXXXXX-XXX-X1 and the NC code is T0000001, and the position (100, 100) is machined using a die, the shape of the die used in the execution machining is 6-60, and the machining method is single-shot. The material of the workpiece is SUS (Steel Use Stainless), and the thickness is 3.5 mm. The processing date and time is 13:00:25 on XX / XX / XX / 20XX.

[0020] Returning to Figure 2, the sensing device 3 measures the machining sound generated when a mold is used in the TPP processing machine 1. The sensing device 3 is equipped with a machining sound acquisition unit 31 that acquires the machining sound of the actual processing from a sound sensor installed near the processing point of the TPP processing machine 1. The machining sound acquisition unit 31 transmits the acquired measurement data of the machining sound of the actual processing to the mold quality determination device 6.

[0021] To determine whether a die is good or bad using measurement data of machining sounds, it is necessary to create threshold data indicating a reference threshold for determining whether the die is good or bad. The threshold data must be set for each machining condition, such as the type of die, the thickness, material, and machining method of the workpiece. Because the TPP machine 1 can accommodate hundreds of types of dies, and the materials and thicknesses of the workpieces vary, creating die quality determination reference data indicating thresholds for all machining conditions, such as those shown in FIG. 4, through experiments would require enormous costs and man-hours. Therefore, the die maintenance system 100 of this embodiment solves this problem by including a die quality determination threshold generation device 5 that automatically generates threshold data indicating a reference threshold for determining whether the die is good or bad.

[0022] The functional configurations of the machining condition setting device 4, the die quality determination threshold generating device 5, and the die quality determination device 6 will be described with reference to Figure 5. The machining condition setting device 4 includes a machining condition acquisition unit 41 that receives input of machining conditions for the TPP machine 1 from the user and generates machining condition data, and a machining condition storage unit 42 that stores the machining condition data. The user inputs the types of dies provided in the TPP machine 1, the machining method for each die, the thickness of the target workpiece, and the material of the target workpiece into the machining condition acquisition unit 41, which generates machining condition data indicating the input contents. The machining condition acquisition unit 41 may acquire machining condition data from an external device or system and store it in the machining condition storage unit 42.

[0023] Here, the machining condition data will be explained using FIG. 6. The example of machining condition data shown in FIG. 6 has the following items: "Die Shape" indicating the shape of the die; "Working Material" indicating the material of the workpiece to be processed by the die; "Working Thickness" indicating the thickness of the workpiece to be processed by the die; and "Processing Method" indicating the machining method of the die. The "Die Shape" item is not limited to the shape of the die, and may be any information that can identify the type of die. For example, if the die shape is 6-60, the workpiece material is SUS, and the thickness is 3.5 mm, the machining method is punching.

[0024] 5, the die quality determination threshold generating device 5 includes an experiment data storage unit 51 that stores experimental data, which is measurement data acquired from experiments conducted in advance, a threshold calculation unit 52 that calculates a threshold that serves as a standard for determining whether the die is good or bad from the experimental data, and a threshold estimation unit 53 that estimates a threshold for determining whether the die is good or bad under machining conditions other than those used in the experiment. The die quality determination threshold generating device 5 is an example of a die quality determination standard generating device. Hereinafter, machining conditions other than those used in the experiment will be referred to as other machining conditions.

[0025] The experimental data storage unit 51 stores, in advance as experimental data, measurement data acquired from the sensing device 3 during processing conducted in advance as experiments for normal molds (hereinafter referred to as normal molds) that should be judged as "good" and molds (hereinafter referred to as abnormal molds) that have become worn and need to be replaced or polished (hereinafter referred to as abnormal molds) that should be judged as "bad."

[0026] The threshold calculation unit 52 uses a statistical analysis method (e.g., principal component analysis) to calculate a threshold for determining whether the mold is good or bad from experimental data of normal and abnormal molds, and stores threshold data indicating the calculated threshold in the threshold memory unit 54.

[0027] The threshold estimation unit 53 estimates thresholds under other conditions based on the threshold data stored in the threshold storage unit 54 by the threshold calculation unit 52, and stores the threshold data indicating the thresholds under other conditions in the threshold storage unit 54. To estimate the thresholds, an algorithm must be created for each processing parameter such as "plate thickness," "material," and "processing method."

[0028] Here, as an example, a method for calculating and estimating the threshold when the plate thickness is different will be described. In the experiment, workpieces of the same material and plate thickness are machined using the same machining method with a normal die and an abnormal die, and measurement data of the machining sounds from each die is obtained. The experimental data storage unit 51 stores the measurement data of the machining sounds from the normal die and the abnormal die obtained in the experiment as experimental data.

[0029] First, the method for calculating the threshold will be described. The threshold calculation unit 52 calculates the maximum value of the sound waveform of the measurement sound data included in the experimental data stored in the experimental data storage unit 51 and performs a process of fast Fourier transforming the measurement sound data. Hereinafter, the process of fast Fourier transforming the measurement sound data will be abbreviated as FFT analysis. Figure 7 shows a graph comparing the integration results of FFT analysis of the experimental data. The first axis of the graph in Figure 7 represents the maximum value of the sound waveform, and the second axis represents the integrated value of the FFT analysis. The graph shows a clear difference between a normal die and an abnormal die. The abnormal die tends to have larger values ​​on both the first and second axes compared to a normal die. This is because the machining noise increases due to increased friction with the workpiece caused by advanced wear on the edge of the die. The threshold calculation unit 52 calculates a threshold for determining whether the die is good or bad based on the difference in experimental data between the normal die and the abnormal die. For example, the maximum value of the sum of the values ​​on the first and second axes of the graph for a normal mold as shown in FIG. 7 may be used as the threshold, or the threshold calculation unit 52 may output a graph comparing the integration results of FFT analysis as shown in FIG. 7, and the user may input the threshold into the threshold calculation unit 52 by looking at the difference in the experimental data for a normal mold and an abnormal mold on the output graph.

[0030] Next, we will explain how to estimate the threshold value under other conditions. Figure 8 is a graph comparing the integration results of FFT analysis of experimental data obtained by processing workpieces of the same material but with four different thicknesses using the same die. As with the graph in Figure 7, the first axis of the graph in Figure 8 is the maximum value of the sound waveform, and the second axis is the integrated value of the FFT analysis. The graph shows that there is a positive correlation that increases as the plate thickness increases. From Figures 7 and 8, it can be inferred that the threshold value for determining whether the die is good or bad also correlates with the plate thickness.

[0031] For example, the volume of the processing noise during blanking is proportional to the volume of the elastic energy released mainly from the punch frame during breakthrough, which is called the breakthrough phenomenon (see Research and Development Grant AF-93012, 1993). １ and plate thickness t ２ (t １ >t ２ ) the sound energy Q generated when processing workpieces of the same material １and Q ２ is the spring constant k and the plate thickness t １ The elongation value of the frame when the workpiece breaks through is x1, and the elongation value of the frame when the workpiece breaks through is x ２ When the conversion rate from elastic energy to sound energy is α, they are expressed by the following equations 1 and 2, respectively.

[0032]

[0033]

[0034] Plate thickness t １ and t ２ and the frame elongation value x １ and x ２ Relationship with １ :t ２ = x １ :x ２ If it is possible to express the relationship between sound energy and plate thickness as follows, the relationship between sound energy and plate thickness is expressed by the following equation 3.

[0035]

[0036] From equation 3, it can be seen that there is a squared relationship between plate thickness and sound energy. For example, if the plate thickness is doubled, the sound energy is quadrupled. Figure 9 is a graph of the experimental data shown in Figure 8, with the first axis representing plate thickness and the second axis representing the integrated value of the FFT analysis. The graph in Figure 9 shows the maximum, minimum, median, and average value of the integrated value of the FFT analysis for each plate thickness. It can also be seen from the graph in Figure 9 that sound energy increases as the plate thickness increases.

[0037] 8 and 9, an experiment is conducted in advance in which several types of workpieces made of the same material but with different thicknesses are machined using the same die, and based on the experimental data, the threshold estimation unit 53 calculates an estimation formula that reflects, in the threshold, the amount of change in the integral value of the FFT analysis that changes with increasing thickness. The threshold estimation unit 53 estimates thresholds under machining conditions with different thicknesses by substituting the die determination threshold calculated by the threshold calculation unit 52 and the plate thickness value into the calculated estimation formula.

[0038] So far, the calculation method and estimation method of the threshold when the plate thickness is different have been described, but the threshold estimation unit 53 calculates estimation formulas in the same way for parameters other than the plate thickness. Parameters other than the plate thickness include not only the material and processing method, but also, for example, when there is a mold that has the same shape as the mold used in the experiment but a different size, an estimation formula may be calculated in the same way.

[0039] The threshold estimation unit 53 acquires machining condition data from the machining condition storage unit 42 of the machining condition setting device 4. The threshold estimation unit 53 may accept input of machining condition data from a user. In this case, the mold maintenance system 100 may not be provided with the machining condition setting device 4. The threshold storage unit 54 may also store threshold data indicating thresholds under machining conditions used in experiments in advance. In this case, the mold quality determination threshold generation device 5 may not be provided with the experiment data storage unit 51 and the threshold calculation unit 52.

[0040] The threshold estimation unit 53 estimates threshold values ​​for other machining conditions among the machining conditions indicated by the machining condition data using an estimation formula for each parameter based on the threshold data stored in the threshold storage unit 54 by the threshold calculation unit 52 and the machining condition data acquired from the machining condition storage unit 42. The threshold estimation unit 53 stores threshold data indicating threshold values ​​under all other machining conditions indicated by the machining condition data in the threshold storage unit 54. Since the threshold calculation unit 52 has already stored threshold data indicating threshold values ​​under the machining conditions used in the experiment in the threshold storage unit 54, threshold data indicating threshold values ​​under all machining conditions indicated by the machining condition data is stored in the threshold storage unit 54.

[0041] The mold quality determination device 6 includes a processing measurement data generation unit 61 that generates processing measurement data that corresponds measurement data and executed processing data, a processing measurement data storage unit 62 that stores the processing measurement data, a quality determination unit 63 that determines whether the mold is good or bad based on the processing measurement data and threshold data, a determination result storage unit 64 that stores determination result data that indicates the determination result of the mold quality determination, and a determination result output unit 65 that outputs the determination result data.

[0042] The machining measurement data generation unit 61 associates the measurement data acquired from the sensing device 3 with the execution machining data acquired from the NC data monitoring device 2 to generate machining measurement data and stores the data in the machining measurement data storage unit 62. The pass / fail judgment unit 63 acquires threshold data corresponding to the machining measurement data stored in the machining measurement data storage unit 62 from the threshold memory unit 54 of the die pass / fail threshold generation device 5. The pass / fail judgment unit 63 judges the pass / fail of each die used in the execution machining indicated by the machining measurement data stored in the machining measurement data storage unit 62 using the threshold indicated by the threshold data acquired from the die pass / fail threshold generation device 5. For example, the pass / fail judgment unit 63 judges the die as pass if the number of times the volume of the machining sound indicated by the machining measurement data exceeds the threshold does not exceed a predetermined number (n≧1), and judges the die as fail if the number of times exceeds the predetermined number (n≧1). The pass / fail judgment unit 63 stores judgment result data indicating the judgment result in the judgment result storage unit 64. The determination result output unit 65 outputs the determination result data stored in the determination result storage unit 64 .

[0043] The judgment result data includes NC data for the actual machining using the die. The judgment result data may be output, for example, by displaying it on a screen, by sending it to a terminal used by the user, or by sending it to an external device or system. The user and the external device or system determine which of the actual machining operations of the TPP processing machine 1 the judgment result is for using, based on the NC data included in the judgment result data. The information identifying the actual machining is not limited to NC data, and may be, for example, a number that uniquely identifies the actual machining.

[0044] Next, the flow of the die quality determination threshold generation process and die quality determination process executed in the die maintenance system 100 will be described with reference to Figures 10 and 11. The die quality determination threshold generation process shown in Figure 10 starts, for example, when a threshold generation instruction is input to the die quality determination threshold generation device 5. When the threshold generation instruction is input to the die quality determination threshold generation device 5, the threshold calculation unit 52 uses a statistical analysis method (for example, principal component analysis) to calculate a threshold for die quality determination from the experimental data of normal and abnormal dies stored in the experimental data storage unit 51 (step S11), and stores threshold data indicating the calculated threshold in the threshold storage unit 54 (step S12).

[0045] The threshold estimation unit 53 acquires machining condition data from the machining condition storage unit 42 of the machining condition setting device 4 (step S13). The threshold estimation unit 53 calculates an estimation formula for each parameter based on the threshold data stored in the threshold storage unit 54 by the threshold calculation unit 52 and the machining condition data acquired from the machining condition storage unit 42 (step S14). The threshold estimation unit 53 uses the calculated estimation formula to estimate thresholds for determining whether a die is good or bad for other machining conditions indicated by the machining condition data (step S15). The threshold estimation unit 53 stores threshold data indicating thresholds for other machining conditions indicated by the machining condition data in the threshold storage unit 54 (step S16), and then ends the process. Since the threshold calculation unit 52 has already stored threshold data indicating thresholds for the machining conditions used in the experiment in the threshold storage unit 54 in step S12, threshold data indicating thresholds for all machining conditions indicated by the machining condition data is stored in the threshold storage unit 54 in step S16.

[0046] 11 starts, for example, when an instruction to judge the quality of a die is input to the die quality judgment device 6. When the instruction to judge the quality of a die is input to the die quality judgment device 6, the machining measurement data generation unit 61 generates machining measurement data by correlating the measurement data acquired from the sensing device 3 with the actual machining data acquired from the NC data monitoring device 2 (step S21). The machining measurement data generation unit 61 stores the generated machining measurement data in the machining measurement data storage unit 62 (step S22). The quality judgment unit 63 acquires threshold data corresponding to the actual measurement data stored in the machining measurement data storage unit 62 from the threshold storage unit 54 of the die quality judgment threshold generation device 5 (step S23).

[0047] The pass / fail judgment unit 63 performs a die pass / fail judgment for each die used in the executed machining indicated by the machining measurement data stored in the machining measurement data storage unit 62, using the threshold indicated by the threshold data acquired from the die pass / fail judgment threshold generation device 5 (step S24). The pass / fail judgment unit 63 stores judgment result data indicating the judgment result of the die pass / fail judgment in the judgment result storage unit 64 (step S25). The judgment result output unit 65 outputs the judgment result data stored in the judgment result storage unit 64 (step S26), and the processing ends. The judgment result data in step S26 can be output by, for example, displaying it on a screen, sending it to a terminal used by the user, or sending it to an external device or system.

[0048] According to the mold maintenance system 100 of the first embodiment, threshold data serving as a standard for determining whether a mold in a TPP machine is good or bad is automatically generated based on measurement data, and the mold quality of the TPP machine is determined based on NC data, measurement data, and the threshold data. This enables automatic determination of the mold quality of the TPP machine, thereby reducing the cost of setting standards for mold quality determination. Furthermore, by outputting the mold quality determination result data, the user can determine the appropriate timing for mold replacement or polishing, thereby preventing excessive and insufficient maintenance. Furthermore, the mold quality determination result data can be used as a basis for determining when to order a mold, contributing to improved facility operation rates.

[0049] (Embodiment 2) In embodiment 2, if the result of the die quality determination is a "fail" determination, that is, if the die being used in the actual machining is determined to be an abnormal die, countermeasures are taken. The countermeasures are to notify the operator that the die is an abnormal die during manned hours when an operator is present (e.g., daytime). Also, to avoid using the abnormal die during unmanned hours when an operator is not present (e.g., nighttime), the NC program of the TPP processing machine 1 is changed.

[0050] The configuration of a mold maintenance system 200 according to the second embodiment will be described with reference to Fig. 12. The mold maintenance system 200 includes a TPP processing machine 1, an NC data monitoring device 2, a sensing device 3, a processing condition setting device 4, a mold quality determination threshold value generating device 5, a mold quality determination device 6, and an NC program changing device 7 that changes the NC program of the TPP processing machine 1.

[0051] The functional configuration of the NC program modification device 7 will be described with reference to Fig. 13. The NC program modification device 7 includes a determination result storage unit 71 that stores determination result data received from the die quality determination device 6, an NC program change command unit 72 that instructs changes to the NC program based on the determination result data, a die replacement master storage unit 73 that indicates the die shape and the combination of replaceable dies for each die, a same-shape die replacement execution unit 74 that replaces an abnormal die in the NC program with a normal die of the same shape, a different-shape die replacement execution unit 75 that replaces an abnormal die in the NC program with a normal die of a different shape, a schedule modification unit 76 that postpones the machining date and time when the abnormal die in the NC program is used, and an NC program storage unit 77 that stores the NC program for the TPP machining machine 1. In the second embodiment, the TPP machining machine 1 performs machining according to the NC program stored in the NC program storage unit 77.

[0052] When the die quality determination device 6 outputs a "fail" determination result, i.e., determination result data indicating that the die is abnormal, and the determination result storage unit 71 stores the result, the NC program change command unit 72 first determines whether the current time is during a manned time period. If it is during a manned time period, the NC program change command unit 72 outputs warning data that warns that a defective die is being used. The warning data may be output, for example, by displaying it on a screen or by sending it to a terminal used by the worker.

[0053] If the current time is an unmanned time slot, the NC program change command unit 72 determines whether or not there is a normal mold of the same shape as the abnormal mold based on the judgment result data stored in the judgment result memory unit 71. If it is determined that there is a normal mold of the same shape, the NC program change command unit 72 causes the same-shape mold replacement execution unit 74 to replace the abnormal mold in the NC program with a normal mold of the same shape. The same-shape mold replacement execution unit 74 changes the NC programs stored in the NC program memory unit 77 that use the abnormal mold to programs that use a normal mold of the same shape. If it is determined that there is no normal mold of the same shape, the NC program change command unit 72 refers to the mold replacement master stored in the mold replacement master memory unit 73 and determines whether or not there is a normal mold of a different shape that can replace the abnormal mold.

[0054] Here, the die replacement master will be explained using Figure 14. In the example of Figure 14, the die replacement master has the following items: "Die Shape" indicating the shape of the die, "Processing Method" indicating the processing method of the die, and "Replacement Die" indicating a die of a different shape that can be substituted for the die. For example, a die with a die shape of 6-110 can be substituted for a die with a die shape of 6-60 in the single-shot and overtaking processing methods.

[0055] 13 , when it is determined that a replaceable normal die of a different shape is available, the NC program change command unit 72 causes the different-shape die replacement execution unit 75 to replace the abnormal die in the NC program with a replaceable normal die of a different shape. The different-shape die replacement execution unit 75 changes the NC programs stored in the NC program storage unit 77 that use the abnormal die to programs that use a replaceable normal die of a different shape.

[0056] If there is no replaceable normal die of a different shape, the NC program change command unit 72 causes the schedule change unit 76 to postpone machining of the NC program that uses the abnormal die. The schedule change unit 76 delays the machining date and time of the NC program that uses the abnormal die, among the NC programs stored in the NC program storage unit 77, for a certain period of time. The period for delaying the machining date and time may be set in advance or may be set by the user. The functions of the other devices are the same as those in the first embodiment.

[0057] Next, the flow of the NC program change process executed by the NC program change device 7 will be described with reference to FIG. 15 . The NC program change process shown in FIG. 15 starts, for example, when determination result data indicating a malfunctioning mold is stored in the determination result storage unit 71. When the determination result data indicating a malfunctioning mold is stored in the determination result storage unit 71, the NC program change command unit 72 determines whether the current time is during a manned time period (step S31). If it is during a manned time period (step S31; YES), the NC program change command unit 72 outputs warning data that a malfunctioning mold is being used (step S32), and the process ends. The warning data may be output, for example, by screen display or by transmission to a terminal used by the operator.

[0058] If the current time is not a manned time zone, i.e., if the current time is an unmanned time zone (step S31; NO), the NC program change command unit 72 determines whether a normal mold of the same shape as the abnormal mold is available based on the judgment result data stored in the judgment result storage unit 71 (step S33). If it is determined that a normal mold of the same shape is available (step S33; YES), the NC program change command unit 72 instructs the same-shape mold replacement execution unit 74 to replace the abnormal mold in the NC program with a normal mold of the same shape. The same-shape mold replacement execution unit 74 replaces the NC program stored in the NC program storage unit 77 that uses the abnormal mold with a program that uses a normal mold of the same shape (step S34), and then terminates the process. If it is determined that a normal mold of the same shape is not available (step S33; NO), the NC program change command unit 72 refers to the mold replacement master stored in the mold replacement master storage unit 73 and determines whether a normal mold of a different shape is available to replace the abnormal mold (step S35).

[0059] In the example of Figure 14, the die replacement master has the following items: "Die Shape" indicating the shape of the die, "Processing Method" indicating the processing method of the die, and "Replacement Die" indicating a die of a different shape that can be replaced with the die. For example, a die with a die shape of 6-110 can be replaced with a die with a die shape of 6-60 in the single-shot and overtaking processing methods.

[0060] 15, if it is determined that there is a replaceable normal die of a different shape (step S35; YES), the NC program change command unit 72 causes the different-shape die replacement execution unit 75 to replace the abnormal die in the NC program with a replaceable normal die of a different shape. The different-shape die replacement execution unit 75 changes the NC programs stored in the NC program storage unit 77 that use the abnormal die to a program that uses a replaceable normal die of a different shape (step S36), and ends the process.

[0061] If there is no normal die that can replace the abnormal die (step S35; NO), the NC program change command unit 72 instructs the schedule change unit 76 to postpone the machining of the NC program that uses the abnormal die. The schedule change unit 76 delays the machining date and time of the NC program that uses the abnormal die among the NC programs stored in the NC program storage unit 77 for a certain period of time (step S37), and ends the process. The NC program change device 7 repeats the NC program change process each time it acquires judgment result data indicating that the die is abnormal. Note that the process of steps S33 to S37 may be executed regardless of whether the current time is during a manned time zone. In this case, steps S31 and S32 can be omitted.

[0062] According to the mold maintenance system 200 of the second embodiment, threshold data serving as a standard for determining whether a mold in a TPP machine is good or bad is automatically generated based on measurement data, and the mold quality of the TPP machine is determined based on NC data, measurement data, and the threshold data. This enables automatic mold quality determination for the TPP machine, thereby reducing the cost of establishing standards for mold quality determination. Furthermore, by taking the above-mentioned countermeasures based on the mold quality determination result data, workers can know when to replace or polish the mold, preventing excessive and insufficient maintenance. Furthermore, the use of abnormal molds can be avoided even during times when workers are not present, preventing significant losses due to poor quality and equipment downtime due to mold damage.

[0063] (Embodiment 3) In embodiments 1 and 2, in order to generate threshold data that serves as a reference for determining whether a die is good or bad, it is necessary to prepare normal and abnormal dies, conduct experiments, and store the experimental data. However, in actual sheet metal processing sites, since TPP processing machines that are already in operation are the starting point of the sheet metal processing, it may be difficult to stop production and conduct experiments. Therefore, in embodiment 3, a trained model that has learned to determine whether a die is good or bad is used instead of threshold data that serves as a reference for determining whether a die is good or bad.

[0064] The configuration of a mold maintenance system 300 according to the third embodiment will be described with reference to Fig. 16. The mold maintenance system 300 includes a TPP processing machine 1, an NC data monitoring device 2, a sensing device 3, a mold quality determination device 6, and a learning device 8 that learns mold quality determination.

[0065] The functional configuration of the die quality determination device 6 and the learning device 8 will be described with reference to Fig. 17. The learning device 8 includes a learning data generation unit 81 that generates learning data from measurement data of machining sounds acquired from the sensing device 3, a model generation unit 82 that learns die quality determination based on the learning data and generates a learned model, and a learned model storage unit 83 that stores the learned model. The learning device 8 is an example of a die quality determination standard generation device.

[0066] The trained model generated by the model generation unit 82 is a model for classifying when the mold is not worn (hereinafter referred to as a "good" judgment) and when the mold is worn (hereinafter referred to as a "bad" judgment). A case where the K-means method, which is unsupervised learning, is applied as the learning algorithm used by the model generation unit 82 will be described. The model generation unit 82 performs clustering using the K-means method on the learning data, and learns to judge whether the mold is good or bad through so-called unsupervised learning.

[0067] As a method for evaluating the reliability of clustering, for example, Pseudo F, which can take into account the cohesiveness within a cluster and the discreteness between clusters, is used. For example, FIG. 18 shows the results of clustering the experimental data shown in FIG. 7 using the K-means method without labels. When Pseudo F is applied using the results shown in FIG. 18 as an example, the data set is the integrated value of the FFT analysis of the mold processing sound and the maximum value of the sound waveform. There are two clusters: a pass / fail judgment, and the cluster with the larger variance is the fail judgment cluster. The Jaccard coefficient between the pass / fail judgment cluster is calculated to be 0.0833. The closer the Jaccard coefficient is to 0, the lower the similarity between the clusters, indicating that the pass / fail classification is successful.

[0068] The learning data generation unit 81 generates learning data that is a combination of the integral value of the FFT analysis of the die machining sound and the maximum value of the sound waveform from the measurement data acquired from the sensing device 3. The model generation unit 82 learns die quality judgment by unsupervised learning in accordance with the learning data generated by the learning data generation unit 81, and generates a trained model. The trained model storage unit 83 stores the trained model generated by the model generation unit 82.

[0069] The die quality determination device 6 includes an inference unit 66 instead of the quality determination unit 63. The inference unit 66 determines whether the die used in the machining indicated by the machining measurement data is good or bad based on the machining measurement data and the learned model stored in the learned model storage unit 83 of the learning device 8. Specifically, the inference unit 66 generates a data set of the integral value of the FFT analysis of the die machining sound and the maximum value of the sound waveform from the measurement data included in the machining measurement data, and inputs this data into the learned model. The learned model outputs a determination result indicating whether the die belongs to a cluster of good or bad determinations. If the die belongs to the cluster of good determinations, the inference unit 66 determines that the die used in the actual machining indicated by the machining measurement data is good. On the other hand, if the die belongs to the cluster of bad determinations, the inference unit 66 determines that the die used in the actual machining indicated by the machining measurement data is bad. The inference unit 66 stores determination result data indicating the result of the die quality determination in the determination result storage unit 64. The functions of the other devices are the same as those of the first and second embodiments.

[0070] Next, the flow of the trained model generation process executed by the learning device 8 will be described with reference to FIG. 19 . The trained model generation process shown in FIG. 19 starts, for example, when a learning instruction for die quality determination is input to the learning device 8. When the learning instruction for die quality determination is input to the learning device 8, the training data generation unit 81 acquires measurement data from the sensing device 3 (step S41). The training data generation unit 81 generates training data, which is a combination of the integrated value of the FFT analysis of the die processing sound and the maximum value of the sound waveform, from the measurement data acquired from the sensing device 3 (step S42). The model generation unit 82 learns die quality determination through unsupervised learning in accordance with the training data generated by the training data generation unit 81, and generates a trained model (step S43). The trained model storage unit 83 stores the trained model generated by the model generation unit 82 (step S44), and the process ends. Note that the measurement data acquired in step S41 may be accumulated and stored. As a result, when the trained model generation process is repeated, the amount of training data increases and the accuracy of the trained model improves.

[0071] According to the mold maintenance system 300 of embodiment 3, a trained model serving as a standard for determining whether a TPP machine mold is good or bad is automatically generated based on measurement data. The quality of the TPP machine mold is then determined based on NC data, measurement data, and the trained model. This enables automatic mold quality determination for the TPP machine, thereby reducing the cost of establishing standards for mold quality determination. Furthermore, it is possible to create threshold data for detecting mold abnormalities without preparing worn molds and conducting experiments to calculate thresholds, contributing to improved equipment utilization rates. Furthermore, by using a trained model that has learned mold quality determination as the standard for mold quality determination, mold quality determination becomes possible even at sheet metal processing sites where experimental data for the TPP machine 1 has not been obtained in advance and the TPP machine 1 is already in operation, without having to stop production and conduct experiments. Furthermore, by repeatedly performing the trained model generation process, the accuracy of the trained model is improved.

[0072] In the above-described first to third embodiments, the mold maintenance system 100, the mold maintenance system 200, and the mold maintenance system 300 each include one TPP processing machine 1, one NC data monitoring device 2, and one sensing device 3, but this is not limited to this. A plurality of sets of the TPP processing machine 1, the NC data monitoring device 2, and the sensing device 3 may be provided. An example of the configuration of a mold maintenance system 400 when this modification is applied to the first embodiment is shown in FIG. 20.

[0073] The mold maintenance system 400 includes TPP machines 1-1 to 1-n, NC data monitoring devices 2-1 to 2-n, and sensing devices 3-1 to 3-n. The NC data monitoring devices 2-1 to 2-n each transmit execution machining data to the mold quality determination device 6 via a data transfer device 9. The sensing devices 3-1 to 3-n each transmit measurement data to the mold quality determination device 6 via the data transfer device 9. By transmitting execution machining data and measurement data from multiple NC data monitoring devices 2 and sensing devices 3 to the mold quality determination device 6 via the data transfer device 9, the mold quality determination of multiple TPP machines can be performed in a unified manner using a single mold quality determination device 6. This modification can also be applied to embodiments 2 and 3.

[0074] In the above-described first to third embodiments, we focused on the characteristic that as the die wear progresses, friction with the workpiece increases, resulting in louder noise during processing. We performed die quality assessment based on measurement data of processing noise acquired by a sound sensor. However, in sheet metal processing sites, various processes other than the TPP machine processing, such as welding and cutting, exist, creating a great deal of disturbing noise. Therefore, even if a highly directional microphone is used as a sound sensor, it is difficult to completely block out noise. Therefore, the sensing device 3 may use, in addition to a sound sensor, contact-type sensors that can be directly attached to the TPP machine 1, such as vibration sensors and load sensors. While sound sensors have difficulty distinguishing between processing noise caused by the target and processing noise caused by disturbances, contact-type sensors can reliably sense the processing phenomenon of the target and obtain highly accurate data. Figure 21 shows a modified example of the sensing device 3 equipped with a vibration sensor in addition to a sound sensor.

[0075] The machining vibration acquisition unit 32 acquires the machining vibration of the actual machining from, for example, a uniaxial vibration sensor attached to the ram mechanism of the TPP machine 1. The machining vibration acquisition unit 32 transmits the acquired measurement data of the machining vibration of the actual machining to the die quality determination device 6. The quality determination unit 63 detects the timing of machining from the measurement data of the machining vibration of the actual machining, and samples the measurement data of the machining sound in synchronization with the machining timing. The learning data generation unit 81 generates learning data from the measurement data of the machining sound and machining vibration acquired from the sensing device 3.

[0076] In the above-described first to third embodiments, the quality of the die is determined based on the machining measurement data, including measurement data of the machining sound acquired by the sound sensor. However, as mentioned above, there are many disturbing noises in the sheet metal processing site. Therefore, to improve the accuracy of the die quality determination, the die quality determination device 6 may take into account the number of punches for each die when determining the quality of the die. In this case, the quality determination unit 63 may calculate the number of punches for each die based on the NC data included in the machining measurement data, or may calculate the number of punches for each die based on the measurement data included in the machining measurement data. Alternatively, punch number data indicating the number of punches for each die may be obtained directly from the TPP processing machine 1.

[0077] The pass / fail judgment unit 63 performs die pass / fail judgment by weighting the threshold judgment so that the greater the number of punches in the die, the heavier the weighting becomes. For example, when die pass / fail judgment is performed using measurement data of machining sounds as in embodiments 1 and 2, and a die is judged to be pass / fail if the number of times the magnitude of the machining sounds exceeds the threshold exceeds a predetermined number (n≧1), the predetermined number (n≧1) is decreased as the number of punches increases. Alternatively, when a trained model for die pass / fail judgment is generated using the K-means algorithm as in embodiment 3, and a cluster is trained as capable of die pass / fail judgment when the Jaccard coefficient is equal to or less than a predetermined value A, the predetermined value A may be decreased as the number of punches increases. Furthermore, in combination with the aforementioned modified example, the pass / fail judgment unit 63 may further take into account measurement data of machining vibrations during the actual machining when judging the die pass / fail. The learning data generation unit 81 may generate learning data from measurement data of machining sounds and machining vibrations and punch count data acquired from the sensing device 3.

[0078] The mold maintenance system 200 of the second embodiment adds an NC program change device 7 to the mold maintenance system 100 of the first embodiment, but the NC program change device 7 may also be added to the mold maintenance system 300 of the third embodiment.

[0079] The hardware configuration of the die quality determination threshold value generating device 5, die quality determination device 6, NC program changing device 7, and learning device 8 will be described with reference to Fig. 22. As shown in Fig. 22, the die quality determination device 6, NC program changing device 7, and learning device 8 each include a temporary storage unit 101, a storage unit 102, a calculation unit 103, an input unit 104, a transmission / reception unit 105, and a display unit 106. The temporary storage unit 101, the storage unit 102, the input unit 104, the transmission / reception unit 105, and the display unit 106 are all connected to the calculation unit 103 via a BUS.

[0080] The calculation unit 103 is, for example, a CPU (Central Processing Unit), and executes the processes of the threshold calculation unit 52 and threshold estimation unit 53 of the die quality determination threshold generation device 5, the machining measurement data generation unit 61, quality determination unit 63 and determination result output unit 65 of the die quality determination device 6, the NC program change command unit 72, same-shape die replacement execution unit 74, different-shape die replacement execution unit 75 and schedule change unit 76 of the NC program change device 7, and the learning data generation unit 81 and model generation unit 82 of the learning device 8 in accordance with the control program stored in the storage unit 102.

[0081] The temporary storage unit 101 is, for example, a random-access memory (RAM). The temporary storage unit 101 loads the control program stored in the storage unit 102 and is used as a work area for the calculation unit 103.

[0082] The storage unit 102 is a non-volatile memory such as a flash memory, a hard disk, a DVD-RAM (Digital Versatile Disc - Random Access Memory), or a DVD-RW (Digital Versatile Disc - Rewritable). The storage unit 102 pre-stores programs for causing the calculation unit 103 to perform the processes of the die quality determination threshold generation unit 5, the die quality determination unit 6, the NC program modification unit 7, and the learning unit 8. The storage unit 102 also supplies data stored by these programs to the calculation unit 103 in accordance with instructions from the calculation unit 103, and stores data supplied from the calculation unit 103. The storage unit 102 includes the experimental data storage unit 51 of the die quality determination threshold generation unit 5, the machining measurement data storage unit 62 and the judgment result storage unit 64 of the die quality determination unit 6, the judgment result storage unit 71, the die replacement master storage unit 73, and the NC program storage unit 77 of the NC program modification unit 7, and the learned model storage unit 83 of the learning unit 8.

[0083] In addition, the experimental data memory unit 51 of the mold pass / fail judgment threshold generation device 5, the processing measurement data memory unit 62 and judgment result memory unit 64 of the mold pass / fail judgment device 6, the judgment result memory unit 71 of the NC program modification device 7, the mold replacement master memory unit 73 and NC program memory unit 77, and the learned model memory unit 83 of the learning device 8 may be provided by an external device or system.

[0084] The input unit 104 is an interface device that connects input devices such as a keyboard, a pointing device, and a voice input device to the BUS. Information input by the user is supplied to the calculation unit 103 via the input unit 104. When the threshold estimation unit 53 of the die quality determination threshold generation device 5 accepts input of processing condition data from the user, the input unit 104 functions as the threshold estimation unit 53.

[0085] The transmitting / receiving unit 105 is a network termination device or a wireless communication device connected to the network, and a serial interface or a LAN (Local Area Network) interface connected thereto. The transmitting / receiving unit 105 functions as the threshold estimation unit 53 of the die quality determination threshold generation device 5, the machining measurement data generation unit 61 and the determination result output unit 65 of the die quality determination device 6, and the learning data generation unit 81 of the learning device 8.

[0086] The display unit 106 is a display device such as an LCD (Liquid Crystal Display), an organic EL (Electroluminescence) display, etc. In a configuration in which the judgment result output unit 65 of the mold quality determination device 6 outputs judgment result data by screen display, the display unit 106 functions as the judgment result output unit 65.

[0087] The processing of the experimental data memory unit 51, threshold calculation unit 52 and threshold estimation unit 53 of the die pass / fail judgment threshold generation device 5, the machining measurement data generation unit 61, machining measurement data memory unit 62, pass / fail judgment unit 63, judgment result memory unit 64 and judgment result output unit 65 of the die pass / fail judgment device 6, the judgment result memory unit 71, NC program change command unit 72, die replacement master memory unit 73, same-shape die replacement execution unit 74, different-shape die replacement execution unit 75, schedule change unit 76 and NC program memory unit 77 of the NC program change device 7, and the learning data generation unit 81, model generation unit 82 and learned model memory unit 83 of the learning device 8 shown in Figures 5, 13 and 17 are executed by a control program using the temporary memory unit 101, calculation unit 103, memory unit 102, input unit 104, transmission / reception unit 105, display unit 106, etc. as resources.

[0088] Furthermore, the above hardware configuration and flowchart are merely examples and can be changed and modified as desired.

[0089] The core components of the die quality determination threshold generating device 5, die quality determination device 6, NC program modifying device 7, and learning device 8, such as the calculation unit 103, temporary storage unit 101, storage unit 102, input unit 104, transmission / reception unit 105, and display unit 106, can be realized using an ordinary computer system rather than a dedicated system. For example, a computer program for executing the above operations may be stored and distributed on a computer-readable recording medium such as a flexible disk, CD-ROM (Compact Disc - Read Only Memory), or DVD-ROM (Digital Versatile Disc - Read Only Memory), and the die quality determination threshold generating device 5, die quality determination device 6, NC program modifying device 7, and learning device 8 may be configured by installing the computer program on a computer. Alternatively, the die quality determination threshold generating device 5, die quality determination device 6, NC program modifying device 7, and learning device 8 may be configured by storing the computer program in a storage device of a server device on a communication network, such as the Internet, and downloading it to an ordinary computer system.

[0090] In addition, when the functions of the die pass / fail judgment threshold generation device 5, die pass / fail judgment device 6, NC program modification device 7, and learning device 8 are realized by sharing the functions between an OS (Operating System) and an application program, or by collaboration between an OS and an application program, only the application program portion may be stored on a recording medium or storage device.

[0091] It is also possible to superimpose a computer program on a carrier wave and provide it via a communication network. For example, the computer program may be posted on a bulletin board system (BBS) on the communication network and provided via the communication network. The computer program may then be started and executed under the control of an OS in the same way as other application programs, thereby enabling the above-mentioned processing to be performed.

[0092] It should be noted that the present disclosure allows various embodiments and modifications without departing from the broad spirit and scope of the present disclosure. Furthermore, the above-described embodiments are intended to explain the disclosure and do not limit the scope of the present disclosure. That is, the scope of the present disclosure is defined by the claims, not the embodiments. Various modifications made within the scope of the claims and within the meaning of the disclosure equivalent thereto are considered to be within the scope of the present disclosure.

[0093] 1, 1-1 to 1-n TPP processing machine, 2, 2-1 to 2-n NC data monitoring device, 3, 3-1 to 3-n sensing device, 4 processing condition setting device, 5 die quality judgment threshold value generation device, 6 die quality judgment device, 7 NC program change device, 8 learning device, 9 data transfer device, 11 press processing unit, 12 NC data display unit, 21 NC data acquisition unit, 22 NC data storage unit, 23 die NC code master storage unit, 24 execution processing data generation unit, 31 processing sound acquisition unit, 32 processing vibration acquisition unit, 41 processing condition acquisition unit, 42 processing condition storage unit, 51 experimental data storage unit, 52 threshold value calculation unit, 53 threshold value estimation unit, 54 threshold value storage unit, 61 processing measurement data generation unit, 62 processing measurement data storage unit, 63 quality judgment unit, 64 judgment result storage unit, 65 Judgment result output unit, 66: Inference unit, 71: Judgment result memory unit, 72: NC program change command unit, 73: Mold replacement master memory unit, 74: Same shape mold replacement execution unit, 75: Different shape mold replacement execution unit, 76: Schedule change unit, 77: NC program memory unit, 81: Learning data generation unit, 82: Model generation unit, 83: Learned model memory unit, 100, 200, 300, 400: Mold maintenance system, 101: Temporary memory unit, 102: Memory unit, 103: Calculation unit, 104: Input unit, 105: Transmission / reception unit, 106: Display unit.

Claims

1. A mold maintenance system comprising: a TPP processing machine which is equipped with a plurality of molds and which selectively uses the molds in accordance with an NC program to perform processing; an NC data monitoring device which acquires NC data which indicates the processing content of the TPP processing machine; a sensing device which generates measurement data which measures processing sounds generated when the molds are used on the TPP processing machine; a mold quality determination standard generating device which generates threshold data or a learned model which serves as the standard for determining whether the molds of the TPP processing machine are good or bad based on the measurement data; and a mold quality determination device which determines whether the molds of the TPP processing machine are good or bad based on the NC data, the measurement data, and the threshold data or the learned model which serves as the standard for determining whether the molds are good or bad.

2. The mold maintenance system of claim 1, wherein the NC data monitoring device comprises: an NC data acquisition unit which reads in the NC data; and an execution machining data generation unit which generates actual machining data by adding data on the mold used in machining to the NC data; the mold pass / fail judgment standard generation device comprises: a threshold calculation unit which calculates a threshold value which is a standard for mold pass / fail judgment from experimental data, which is the measurement data acquired in a previously conducted experiment, and generates threshold data which indicates the calculated threshold value; and a threshold estimation unit which estimates a threshold value for mold pass / fail judgment under machining conditions other than the machining conditions used in the experiment based on the threshold value calculated by the threshold calculation unit, and generates threshold data which indicates the estimated threshold value; and the mold pass / fail judgment device comprises: a machining measurement data generation unit which generates machining measurement data which corresponds to the measurement data and the execution machining data; and a pass / fail judgment unit which judges the pass / fail of the mold for the TPP processing machine based on the machining measurement data and the threshold data.

3. A mold maintenance system as described in claim 2, further comprising a processing condition setting device that accepts input of processing conditions for the TPP processing machine, wherein the threshold estimation unit estimates a threshold for determining whether the mold is good or bad under processing conditions other than the processing conditions used during the experiment, among the processing conditions input to the processing condition setting device.

4. The mold maintenance system described in claim 1, wherein the NC data monitoring device comprises: an NC data acquisition unit that reads the NC data; and an execution machining data generation unit that generates actual machining data by adding data on the mold used in machining to the NC data; the mold quality determination standard generation device comprises: a learning data generation unit that generates learning data from the measurement data; and a model generation unit that learns the mold quality determination for the TPP processing machine based on the learning data and generates a learned model; and the mold quality determination device comprises: a machining measurement data generation unit that generates machining measurement data that corresponds the measurement data and the execution machining data; and an inference unit that performs quality determination for the mold of the TPP processing machine based on the machining measurement data and the learned model.

5. A mold maintenance system as described in any one of claims 2 to 4, comprising a plurality of the TPP processing machines, a plurality of the NC data monitoring devices, and a plurality of the NC data monitoring devices, each of the plurality of NC data monitoring devices transmitting the executed processing data to the mold quality determination device via a data transfer device, and each of the plurality of sensing devices transmitting the measurement data to the mold quality determination device via the data transfer device.

6. An NC program change device is further provided which changes the NC program of the TPP processing machine based on the judgment result of the die quality judgment device, the die quality judgment device outputs judgment result data indicating the judgment result of the die quality judgment of the TPP processing machine to the NC program change device, the NC program change device judges whether there is a normal die having the same shape as the abnormal die when the judgment result data indicates that the die is abnormal, and if it is judged that there is a normal die having the same shape, replaces the abnormal die in the NC program with a normal die having the same shape, and if it is judged that there is no normal die having the same shape as the abnormal die, judges whether there is a normal die having a different shape that can be replaced with the abnormal die, and if it is judged that there is a replaceable normal die having a different shape, replaces the abnormal die in the NC program with a replaceable normal die having a different shape, and if it is judged that there is no replaceable normal die having a different shape, postpones the processing of the NC program using the abnormal die; The mold maintenance system according to any one of claims 1 to 5, comprising: a same-shape mold replacement execution unit that replaces the abnormal mold in the NC program with a normal mold of the same shape in accordance with instructions from the NC program change command unit; a different-shape mold replacement execution unit that replaces the abnormal mold in the NC program with a normal mold of a different shape in accordance with instructions from the NC program change command unit; and a schedule change unit that postpones the processing date and time when the abnormal mold in the NC program is used in accordance with instructions from the NC program change command unit.

7. The mold maintenance system described in claim 6, wherein the NC program change command unit further determines whether the current time is a manned time zone, and if so, outputs warning data to warn that an abnormal mold is being used, and if so, determines whether a normal mold of the same shape as the abnormal mold is available, and if so, replaces the abnormal mold in the NC program with a normal mold of the same shape, and if so, determines whether a normal mold of the same shape as the abnormal mold is not available, determines whether a normal mold of a different shape that can be replaced with the abnormal mold is available, and if so, replaces the abnormal mold in the NC program with a replaceable normal mold of a different shape, and if so, postpones processing of the NC program which uses the abnormal mold, 8. A die quality determination threshold generation device comprising: a threshold calculation unit that calculates a threshold value that serves as a standard for determining whether a die in a TPP processing machine is good or bad from experimental data obtained by measuring processing sounds generated when a die is used in the TPP processing machine in a previously conducted experiment, and generates threshold data indicating the calculated threshold value; and a threshold estimation unit that estimates a threshold value for determining whether a die is good or bad under processing conditions other than the processing conditions used in the experiment, based on the threshold value calculated by the threshold calculation unit, and generates the threshold data indicating the estimated threshold value.

9. A learning device comprising: a learning data generation unit that generates learning data from measurement data obtained by measuring machining sounds generated when a mold is used in a TPP processing machine; and a model generation unit that learns how to judge the quality of the molds of the TPP processing machine based on the learning data, and generates a learned model.

10. A method for determining whether a die is good or bad, comprising: a step executed by an NC data monitoring device to acquire NC data indicating the processing contents of a TPP processing machine which has a plurality of dies and selectively uses the dies in accordance with an NC program to perform processing; a step executed by a sensing device to generate measurement data measuring processing sounds generated when the dies are used on the TPP processing machine; a step to generate threshold data or a learned model based on the measurement data, which serves as a standard for determining whether the dies of the TPP processing machine are good or bad; and a step to determine whether the dies of the TPP processing machine are good or bad, based on the NC data, the measurement data, and the threshold data or the learned model, which serves as a standard for determining whether the dies are good or bad.

11. A program that causes a computer to function as: a threshold calculation unit that calculates a threshold value that serves as a standard for determining whether a die in a TPP processing machine is good or bad from experimental data measuring the processing sounds generated when a die is used in the TPP processing machine in a previously conducted experiment, and generates threshold data indicating the calculated threshold value; and a threshold estimation unit that estimates a threshold value for determining whether a die is good or bad under processing conditions other than the processing conditions used in the experiment, based on the threshold value calculated by the threshold calculation unit, and generates the threshold data indicating the estimated threshold value.

12. A program that causes a computer to function as: a learning data generation unit that generates learning data from measurement data obtained by measuring the machining sounds generated when a mold is used in a TPP processing machine; and a model generation unit that learns how to judge the quality of molds in the TPP processing machine based on the learning data, and generates a learned model.

Images