Diagnostic system, diagnostic method, and diagnostic program

Abstract

The diagnostic system (1) includes a machining data acquisition unit (10) that acquires machining data from the start of machining by a machine tool (2) to the end of machining, a section extraction unit (11) that extracts a first section, which is a cutting section, and a second section, which is a section other than the first section, from the machining sections of the machining data, a feature calculation unit (13) that calculates a first feature value in the first section of the machining data and a second feature value in the second section of the machining data, a corrected feature calculation unit (14) that calculates a third feature value corresponding to an actual cutting load, which is a load of the cutting process, by subtracting the second feature value from the first feature value, and a machining diagnosis unit (17) that performs machining diagnosis based on the third feature value.

Description

[Technical Field]

[0001] The present disclosure relates to a diagnostic system, a diagnostic method, and a diagnostic program for diagnosing machining by a machine tool. [Background technology]

[0002] When a machine tool processes a workpiece using a tool, the machining state of the workpiece by the tool changes due to the state of the tool, the state of the workpiece, etc. If machining is performed in an abnormal machining state, the desired machined product cannot be obtained. Furthermore, if a tool is replaced in a normal machining state, the manufacturing cost of the machined product increases. For this reason, it is desirable to accurately perform machining diagnosis in machine tools.

[0003] The machining diagnosis device described in Patent Document 1 extracts and cleanses machining data of a cutting section from machining data such as current values, calculates feature values ​​from the extracted and cleansed machining data, and performs machining diagnosis based on the calculated feature values. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 6949275 Summary of the Invention [Problem to be solved by the invention]

[0005] However, in the technology of Patent Document 1, for example, feature values ​​are calculated for machining data including those caused by machining processing and those caused by external factors other than machining processing, such as temperature changes in the machine tool. Therefore, when there is little change in the feature values ​​of machining data calculated from machining data in cutting processing, the accuracy of machining diagnosis is low.

[0006] The present disclosure has been made in consideration of the above, and aims to provide a diagnostic system that can achieve highly accurate machining diagnosis even when the change in the feature value of machining data calculated from machining data in cutting processing is small. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems and achieve the object, the diagnostic system of the present disclosure includes: a machining data acquisition unit that acquires machining data from the start to the end of machining by a machine tool; The waveform of the pre-processing data, which is the processing data when the air cutting processing and the actual processing are performed in advance, is acquired based on the waveform of the pre-processing data. Processing data waveform From the first section, which is the cutting section The first waveform of and a second section other than the first section. The second waveform of The diagnostic system of the present disclosure further comprises: First Waveform a first feature value in a first section; Second Waveform a feature calculation unit that calculates a second feature in the second section of the cutting work; a corrected feature calculation unit that calculates a third feature corresponding to the actual cutting load, which is the load of the cutting work, by subtracting the second feature from the first feature; and a machining diagnosis unit that performs machining diagnosis based on the third feature. [Effects of the Invention]

[0008] The diagnostic system according to the present disclosure has an effect of being able to realize highly accurate machining diagnosis even when the change in the feature amount of machining data calculated from machining data in cutting is small. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram showing a configuration of a diagnostic system according to an embodiment. [Figure 2] FIG. 10 is a diagram for explaining a waveform of processed data acquired by the diagnostic system according to the embodiment; [Figure 3] FIG. 1 is a diagram for explaining a cutting processing section and a correction section extracted from processing data by a diagnostic system according to an embodiment. [Figure 4]1 is a flowchart showing a procedure of processing executed by a diagnostic system according to an embodiment; [Figure 5] FIG. 10 is a diagram for explaining a process in which the diagnostic system according to the embodiment calculates an actual feature amount using an integral value of processed data. [Figure 6] FIG. 10 is a diagram for explaining a process in which the diagnostic system according to the embodiment determines the machining state and the wear state of the tool based on the feature amount. [Figure 7] FIG. 10 is a diagram for explaining a first processing example in which the diagnostic system according to the embodiment determines the machining state based on the change tendency of the actual feature amount; [Figure 8] FIG. 10 is a diagram for explaining a second processing example in which the diagnostic system according to the embodiment determines the machining state based on the change tendency of the actual feature amount. [Figure 9] FIG. 10 is a diagram for explaining a third processing example in which the diagnosis system according to the embodiment determines the machining state based on the change tendency of the actual feature amount. [Figure 10] FIG. 10 is a diagram for explaining a fourth processing example in which the diagnostic system according to the embodiment determines the machining state based on the change tendency of the actual feature amount. [Figure 11] FIG. 10 is a diagram for explaining a fifth processing example in which the diagnostic system according to the embodiment determines the machining state based on the change tendency of the actual feature amount. [Figure 12] FIG. 10 is a diagram for explaining a sixth processing example in which the diagnostic system according to the embodiment determines the machining state based on the change tendency of the actual feature amount. [Figure 13] FIG. 10 is a diagram for explaining a first processing example in which the diagnostic system according to the embodiment determines the machining state based on the actual feature amount; [Figure 14] FIG. 10 is a diagram for explaining a second processing example in which the diagnostic system according to the embodiment determines the machining state based on the actual feature amount; [Figure 15] FIG. 10 is a diagram for explaining a third processing example in which the diagnosis system according to the embodiment determines the machining state based on the actual feature amount. [Figure 16] FIG. 10 is a diagram for explaining an example of transition of actual feature values ​​calculated by the diagnostic system according to the embodiment when a tool deteriorates. [Figure 17]FIG. 17 is a diagram for explaining a process in which the diagnostic system according to the embodiment determines the machining state in the second processing example for the graph shown in FIG. [Figure 18] FIG. 17 is a diagram for explaining a first example of processing in which the diagnostic system according to the embodiment determines the machining state in the third processing example for the graph shown in FIG. [Figure 19] FIG. 17 is a diagram for explaining a second example of processing in which the diagnostic system according to the embodiment determines the machining state in the third processing example for the graph shown in FIG. [Figure 20] FIG. 10 is a diagram for explaining spindle load data used by a diagnostic system of a comparative example. [Figure 21] FIG. 10 is a diagram for explaining feature amounts used by a diagnostic system of a comparative example. [Figure 22] FIG. 1 is a diagram for explaining the timing of tool replacement when the diagnostic system according to the embodiment determines the wear state of the tool. [Figure 23] FIG. 1 is a diagram illustrating a configuration example of a processing circuit when the processing circuit included in the diagnostic system according to the embodiment is realized by a processor and a memory. [Figure 24] FIG. 1 is a diagram illustrating an example of the configuration of a processing circuit when the processing circuit included in the diagnostic system according to the embodiment is configured with dedicated hardware. DETAILED DESCRIPTION OF THE INVENTION

[0010] A diagnostic system, a diagnostic method, and a diagnostic program according to embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0011] Embodiment Fig. 1 is a diagram showing the configuration of a diagnostic system according to an embodiment. Diagnostic system 1 is a system that diagnoses machining by a machine tool 2. Machining diagnosis performed by diagnostic system 1 includes diagnosis of tool wear conditions, diagnosis of whether the machining state is abnormal, diagnosis of signs of abnormality in machine tool 2, etc.

[0012] Machine tool 2 cuts a workpiece (not shown) that is a workpiece to be processed using a tool. Machine tool 2 is a machine that performs processes such as cutting, cutting, and polishing on the workpiece that is the object to be processed, and is, for example, a milling machine, a lathe, or a drill press. Diagnostic system 1 diagnoses the cutting process performed by machine tool 2 and outputs the diagnosis results to machine tool 2.

[0013] The diagnostic system 1 includes a machining data acquisition unit 10, a section extraction unit 11, a feature calculation unit 13, a corrected feature calculation unit 14, a change trend calculation unit 15, a learning unit 16, a machining diagnosis unit 17, and a diagnosis result output unit 18. The diagnostic system 1 also includes a trend data storage unit 22 and a diagnostic model storage unit 23. The section extraction unit 11 includes a correction section extraction unit 12A and a cutting processing section extraction unit 12B.

[0014] The machining data acquiring unit 10 acquires machining data, machining information used to identify the machining type (machining conditions), and control information (including machining method, etc.) via a CNC (Computer Numerical Control) from the machine tool 2. Note that the machining data acquiring unit 10 may acquire the machining data and machining information directly from the machine tool 2.

[0015] The machining data for one machining process (the process of cutting out one product) includes machining data for the approach section where the tool approaches the workpiece, machining data for the cutting section (cutting section) where the tool cuts the workpiece, and machining data for the retreat section (tool retreat section) where the tool retreats from the workpiece. The cutting section is the first section, and the approach section or the retreat section is the second section.

[0016] The machining data in the cutting processing section is the machining data to be diagnosed. The machining data in the approach section and the machining data in the retreat section are the machining data for correcting the machining data in the cutting processing section. In this way, the cutting processing section is the section to be diagnosed (diagnosis section), and the approach section and the retreat section are sections for correcting the machining data in the cutting processing section (correction section). The correction section is a section other than the cutting processing section.

[0017] Here, the machining data, the cutting processing section, and the correction section will be described. Fig. 2 is a diagram for explaining the waveform of the machining data acquired by the diagnostic system according to the embodiment. Fig. 3 is a diagram for explaining the cutting processing section and the correction section extracted from the machining data by the diagnostic system according to the embodiment.

[0018] The horizontal axis of the graphs shown in Figs. 2 and 3 is time, and the vertical axis is cutting load (spindle load in Figs. 2 and 3). The spindle load corresponds to a physical quantity (such as a motor drive current value) included in the machining data. Fig. 2 shows the waveform Wm of the spindle load as well as the waveform Rm of the spindle rotation speed. Fig. 3 shows the waveform Wm of the spindle load extracted by the diagnostic system 1 from the waveform of the machining data shown in Fig. 2, as well as the cutting section, correction section (approach section, retract section) and the like in the waveform Wm.

[0019] When cutting starts, the tool starts to rotate and the spindle rotation speed increases to the target rotation speed (the rotation speed during actual cutting). In other words, the rotation of the tool accelerates and the rotation speed increases to the target speed (the rotation speed during actual cutting). The section from when the tool starts to rotate until the rotation speed reaches the target speed is the acceleration section IN1.

[0020] When the spindle starts to rotate, an acceleration torque is generated, and the spindle load corresponding to the motor drive current value, etc., increases. As the spindle rotation approaches the target rotation speed, the spindle load decreases and stabilizes. The spindle load when the spindle load is stable is the spindle load required to maintain rotation, and this interval is the approach interval. In other words, the interval from when the spindle rotation speed reaches the rotation speed during actual machining until the tool comes into contact with the workpiece is the approach interval IN2.

[0021] Machine tool 2 starts actual machining by bringing the tool closer to the workpiece and bringing it into contact with the workpiece. When the tool starts cutting the workpiece, cutting resistance is generated by cutting the workpiece, which generates a spindle load due to actual cutting, and the motor drive current value corresponding to the spindle load increases. Note that when machining begins, the loads on axes other than the spindle also increase in the same way as the spindle load. The section where the tool is in contact with the workpiece and cutting is being performed is the actual machining section IN3.

[0022] After that, when cutting of the workpiece is completed, the machine tool 2 moves the tool away from the workpiece. As a result, the spindle load decreases and stabilizes. The interval from when cutting of the workpiece is completed until the spindle load decreases and stabilizes is the retreat interval IN4.

[0023] After this, the rotation speed of the tool decreases and the rotation of the tool stops. The period from when the rotation speed of the tool decreases to when the rotation of the tool stops is the deceleration period IN5. The machining data acquisition unit 10 acquires machining data including machining data of the actual machining section IN3 and machining data of at least one of the approach section IN2 and the retreat section IN4.

[0024] In Figure 2, the spindle load in the approach section or tool retraction section is illustrated as base load Lb. Base load Lb, which is the load in the approach section or tool retraction section, may fluctuate regardless of machining. For example, when the machine tool 2 starts machining in a cold state without being warmed up, the machining data increases overall. Specifically, during machining in a cold state, the base load Lb increases, and the spindle load during cutting also increases by the amount of the increase in base load Lb. In other words, when the machining data increases or decreases due to factors other than machining, the increase or decrease is reflected in the spindle load. In this case, conventionally, an unusual spindle load different from that during normal machining operation is calculated, and the unusual spindle load may be determined to be abnormal.

[0025] Furthermore, in the past, when a significant machining abnormality such as tool breakage occurred, the spindle load changed significantly, making it possible to detect the abnormality, but it was impossible to detect minor tool defects (chipping) or machining abnormalities that only caused a slight change.

[0026] The diagnostic system 1 of this embodiment calculates the spindle load during cutting, eliminating factors other than machining, such as a cold state, by subtracting the base load Lb from the spindle load during cutting. In FIG. 2, the spindle load during cutting, eliminating factors other than machining, is shown as the actual cutting load Lc. In this way, the diagnostic system 1 calculates the actual cutting load Lc by subtracting the base load Lb from the spindle load during cutting, in order to eliminate fluctuations in the spindle load caused by factors other than machining. This enables the diagnostic system 1 to perform a highly accurate diagnosis, eliminating factors other than machining.

[0027] The machining data acquired by the machining data acquisition unit 10 includes physical quantities during machining that are used to calculate the machining state. The physical quantities included in the machining data include the motor drive current value, motor drive voltage value, motor rotation speed, motor torque, cutting resistance value, acceleration (vibration), and strain amount obtained from sensors arranged on the machine tool 2. The strain amount is, for example, the strain amount of the tool, tool holder, workpiece, and workpiece chucking mechanism. The strain amount is acquired by a strain sensor.

[0028] The machining data acquisition unit 10 acquires, as machining data, for example, a waveform of a motor drive current value (drive current waveform), a waveform of a motor drive voltage value (drive voltage waveform), a waveform of a motor acceleration, a waveform of a motor torque, acceleration (vibration), a waveform of a strain amount, etc. Note that the physical quantities included in the machining data may be calculated from information detected by a sensor. The machining data may also include machining command values ​​such as a spindle rotation speed command, inspection results of machining dimensions, cutting oil discharge pressure, cutting oil temperature, etc.

[0029] The machining information includes a machining program number, a subprogram number, a tool number, etc. The machining information may also include a machining program including a numerical control command, a workpiece type number, the number of times the tool has been used, manufacturing information of the tool (such as a manufacturing serial number), and machining conditions (feed rate, cutting depth, and operation control method). The numerical control command is a control command used when the machine tool 2 is numerically controlled, such as G0 (positioning command), G1 (linear interpolation), and G2 (circular interpolation, clockwise).

[0030] The machining data acquiring unit 10 also identifies the machining type based on the machining information. For example, the machining data acquiring unit 10 uniquely identifies the machining type based on a combination of a machining program and a tool number in the machining information or a machining program line.

[0031] The machining data acquiring unit 10 associates the acquired machining data for one machining process with the machining type, and transmits the associated data to the correction section extracting unit 12A and the cutting processing section extracting unit 12B. The diagnostic system 1 may have a machining data storage unit (not shown) that stores the machining data and machining type acquired by the machining data acquiring unit 10. In this case, the machining data storage unit stores the machining data in association with the machining type.

[0032] The diagnostic system 1 automatically distinguishes between the approach section before cutting, the cutting section, and the retreat section after cutting by having the machine tool 2 execute air cutting processing (an operation that does not process the workpiece) and actual processing in advance in the machining program to be diagnosed, and stores each section.

[0033] If the diagnostic system 1 can determine the spindle load (such as the motor drive current value) immediately before and after the actual cutting that occurs due to the actual cutting, it can determine the section in which the actual cutting occurred. For example, when an air cutting operation is performed, machining data is obtained in which the spindle load due to the actual cutting does not occur. The spindle load immediately before and after the spindle load that occurs due to the actual cutting can be considered to be a spindle load that is independent of the actual cutting. For this reason, when an air cutting operation is performed in the same type of machining, the spindle load (machining data) that does not include the spindle load due to the actual cutting is obtained.

[0034] In this embodiment, the machine tool 2 is caused to perform air cutting and actual machining as the same type of machining in advance. Then, the correction section extraction unit 12A of the diagnostic system 1 extracts a cutting section, which is a section where actual cutting is performed, from the difference in spindle load between air cutting and actual machining. That is, the diagnostic system 1 extracts a cutting section where actual cutting is performed from the difference between the spindle load when air cutting is performed and the spindle load when actual machining is performed. Furthermore, the correction section extraction unit 12A determines sections where the spindle load is stable before and after the section where actual cutting is performed, and designates these sections as the approach section and the evacuation section, respectively. Note that the section between the approach section and the evacuation section may also be considered as the cutting section.

[0035] The correction interval extraction unit 12A may store the waveform (amount of change, rate of change, etc.) of the machining data at the timing when each interval starts and ends, or may store the time when each interval starts and ends after the tool starts rotating.

[0036] The correction interval extraction unit 12A stores, for example, for each machining type, an approach waveform which is the waveform of machining data in the approach interval, an evacuation waveform which is the waveform of machining data in the evacuation interval, and a cutting waveform which is the waveform of machining data in the cutting interval. Furthermore, the correction interval extraction unit 12A may automatically determine the approach waveform, the evacuation waveform, and the cutting waveform in intervals such as G0 and G1 of the machining program.

[0037] When diagnosing machining, the correction interval extraction unit 12A extracts machining data of a correction interval (at least one of an approach interval and a retreat interval) from the machining data acquired by the machining data acquisition unit 10. Specifically, the correction interval extraction unit 12A extracts the waveform of the machining data (base load) of the correction interval from the waveform of the machining data (physical quantity) based on the type of machining. That is, the correction interval extraction unit 12A extracts the waveform of the machining data of the approach interval from the machining data based on the approach waveform, and extracts the waveform of the machining data of the retreat interval from the machining data based on the retreat waveform.

[0038] The machining data of the correction section extracted by the correction section extraction unit 12A is data when the tool is not in contact with the workpiece, and is therefore data that does not depend on the machining dimensions and machining time. The correction section extraction unit 12A transmits the machining type and the extracted machining data of the approach section and the retreat section to the feature calculation unit 13.

[0039] When diagnosing machining, the cutting work section extraction unit 12B extracts machining data of the cutting work section from the machining data acquired by the machining data acquisition unit 10. Specifically, the cutting work section extraction unit 12B extracts the waveform of the machining data (actual cutting load) of the cutting work section from the waveform of the machining data based on the machining type. That is, the cutting work section extraction unit 12B extracts the waveform of the machining data of the cutting work section from the machining data based on the cutting work waveform.

[0040] The machining data of the cutting processing section extracted by the cutting processing section extraction unit 12B is data when the tool is in contact with the workpiece, and therefore is data that depends on the machining dimensions and machining time. The cutting processing section extraction unit 12B transmits the machining type and the machining data of the extracted cutting processing section to the feature amount calculation unit 13.

[0041] The diagnostic system 1 may extract the machining data of the correction section and the cutting section using any method. For example, the diagnostic system 1 may extract the machining data of the correction section and the cutting section from the machining data based on a machining program including NC control commands. In this case, the correction section extraction unit 12A determines the approach section and the retreat section based on the machining program, and extracts the machining data of the approach section and the retreat section from the machining data. Furthermore, the cutting section extraction unit 12B determines the cutting section based on the machining program, and extracts the machining data of the cutting section from the machining data.

[0042] For example, if the numerical control command is G0, it can be determined that the movement section is one in which no machining is performed, and if it is G1 or G2, it can be determined that the section includes machining that operates at a constant speed. Therefore, the correction section extraction unit 12A may determine the G0 section as the correction section, and the cutting processing section extraction unit 12B may determine the G1 or G2 section as the cutting processing section.

[0043] The feature amount calculation unit 13 calculates the feature amount of the correction section from the processing data of the correction section, and calculates the feature amount of the cutting processing section from the processing data of the cutting processing section. The feature amount calculation unit 13 calculates the feature amount of the correction section from the processing data of at least one of the approach section and the retreat section. For example, the feature amount calculation unit 13 calculates the average value of the feature amount calculated from the processing data of the approach section and the feature amount calculated from the processing data of the retreat section as the feature amount of the correction section. The feature amount of the correction section is a feature amount for correcting the feature amount of the cutting processing section.

[0044] The feature amount is, for example, a statistical value for one machining operation. Examples of the feature amount are at least one of the average value, maximum value, minimum value, integral value, median value, range value (difference between the maximum and minimum values), and standard deviation of the machining data. The integral value is a value obtained by integrating the feature amount over time. Therefore, the feature amount for the cutting processing section is a value obtained by integrating the machining data over the time of the cutting processing section. Furthermore, the feature amount for the approach section is a value obtained by integrating the machining data over the time of the approach section, and the feature amount for the retreat section is a value obtained by integrating the machining data over the time of the retreat section.

[0045] The feature amount calculation unit 13 transmits the calculated feature amounts to the corrected feature amount calculation unit 14 and the change trend calculation unit 15.

[0046] The corrected feature calculation unit 14 calculates a feature corresponding to actual cutting (actual feature) for the tool currently in use based on the feature of the cutting section and the feature of the correction section. The feature of the cutting section is the first feature, the feature of the correction section is the second feature, and the actual feature is the third feature.

[0047] The corrected feature quantity calculation unit 14 calculates actual feature quantities corresponding to the actual cutting work by performing various arithmetic operations on the feature quantities of the correction section and the feature quantities of the cutting work section. The corrected feature quantity calculation unit 14 calculates the actual feature quantities, for example, by subtracting the feature quantities of the correction section from the feature quantities of the cutting work section. That is, the corrected feature quantity calculation unit 14 corrects the actual feature quantities of the cutting work section by subtracting feature quantities (feature quantities of the approach section or the retreat section) that do not depend on the machining dimensions (machining time) from the feature quantities of the cutting work section. Therefore, the actual feature quantities calculated by the corrected feature quantity calculation unit 14 are feature quantities obtained by correcting the feature quantities of the cutting work section using the feature quantities of the correction section.

[0048] The corrected feature amount calculation unit 14 may calculate the actual feature amount by a process other than subtraction. When the feature amount is an integral value, the corrected feature amount calculation unit 14 calculates the actual feature amount by subtraction according to the ratio between the time of the correction interval and the time of the cutting interval. The calculation process of the actual feature amount when the feature amount is an integral value will be described later.

[0049] When a diagnostic model for diagnosing the machining state, tool wear state, etc. is generated, the corrected feature quantity calculation unit 14 transmits the machining type and the actual feature quantity to the learning unit 16. Furthermore, when the machining state, tool wear state, etc. are actually diagnosed after the diagnostic model is generated, the corrected feature quantity calculation unit 14 associates the machining type with the actual feature quantity and transmits them to the trend data storage unit 22 and the machining diagnosis unit 17.

[0050] The change trend calculation unit 15 calculates the change trend (amount of change, rate of change, etc.) of the feature amount from the past to the present for each machining type. The change trend of the feature amount will be described in detail later. The change trend calculation unit 15 calculates the change trend of the feature amount based on a plurality of machining data acquired over a plurality of machining sections (e.g., 10 machining sections) in which machining was performed under the same machining conditions from the past machining to the present machining. Here, one machining section corresponds to the machining of one workpiece (one machining operation). In other words, one machining section includes one cutting machining section.

[0051] The processing section here may be a section that combines the correction section and the cutting processing section, or may be a cutting processing section. In other words, the processing section may be a processing section in which the cutting processing section and the correction section can be distinguished, or may be a processing section in which they cannot be distinguished.

[0052] In the following, a case will be described in which the processing section used when the change trend calculation unit 15 calculates the change trend of the feature amount is a cutting processing section, but the processing section may include a correction section and a cutting processing section. The change trend calculation unit 15 associates the processing type with the change trend of the feature amount and stores them in the trend data storage unit 22.

[0053] The trend data storage unit 22 receives and stores the actual feature amounts for each machining type from the corrected feature amount calculation unit 14. The trend data storage unit 22 stores the transition of the actual feature amounts for each machining type by receiving and storing the actual feature amounts for each machining from the corrected feature amount calculation unit 14. The trend data storage unit 22 stores the transition of the new actual feature amounts every time the tool is replaced.

[0054] Moreover, the trend data storage unit 22 receives and stores the change trends of the feature amounts for each machining type from the change trend calculation unit 15. The information stored in the trend data storage unit 22 is used by the machining diagnosis unit 17 when diagnosing the machining state and the wear state of the tool.

[0055] The learning unit 16 generates a diagnostic model for each machining type based on past actual feature values. The diagnostic model is a model that diagnoses the machining state and tool wear state based on the actual feature values. The diagnostic model is, for example, a model (threshold determination model) that determines whether the actual feature values ​​exceed a threshold value for the actual feature values. In this case, an upper threshold and a lower threshold value for the actual feature values ​​are set in the diagnostic model through learning. The upper threshold value is a first threshold value, and the lower threshold value is a second threshold value.

[0056] The upper threshold is a threshold of an actual feature such as a motor drive current value for determining whether or not to replace a tool. When the tool wears out and the actual feature becomes equal to or greater than the upper threshold, the tool is replaced. The lower threshold is a threshold of an actual feature for determining whether or not the tool is in an abnormal state (e.g., broken). When the tool becomes abnormal and the actual feature becomes equal to or less than the lower threshold, the tool is replaced.

[0057] The upper and lower thresholds of the diagnostic model do not necessarily have to be generated by learning, but may be set by a user.

[0058] The diagnostic model may also be a model that predicts the life of a tool based on the transition of actual feature values ​​for the tool currently in use. That is, the diagnostic model may predict the life of the tool as the wear state of the tool. In this case, the diagnostic model calculates a regression line based on the transition of actual feature values ​​for multiple machining operations using the tool currently in use, and predicts the life of the tool based on the calculated regression line and an upper limit threshold. The diagnostic model may also predict the life of the tool based on the transition of actual feature values ​​(such as current values ​​and rate of increase) and the upper limit threshold of the diagnostic model.

[0059] The learning unit 16 learns the upper and lower thresholds of the diagnostic model by learning past transitions of the actual feature quantities, and generates the diagnostic model. The learning unit 16 learns the upper and lower thresholds by, for example, multiple regression analysis, decision tree, random forest, etc.

[0060] The learning unit 16 excludes from learning past transitions of actual feature quantities, those transitions of actual feature quantities whose increasing trend is different from others. That is, the learning unit 16 learns the upper and lower thresholds based on past transitions of actual feature quantities whose increasing trend is normal (for example, actual feature quantities whose average value of the rate of increase is within a reference range). Furthermore, the learning unit 16 excludes from learning transitions of actual feature quantities that are equal to or below the reference value when a tool is replaced due to tool wear.

[0061] The learning unit 16 learns the upper and lower thresholds based on, for example, the actual feature values ​​when a tool is replaced with a new tool and the actual feature values ​​when a tool is replaced due to tool wear. The learning unit 16 sets, for example, the minimum value of the actual feature values ​​when a tool is replaced with a new tool as the lower threshold. The learning unit 16 also sets, for example, the maximum value of the actual feature values ​​when a tool is replaced with a new tool due to tool wear as the upper threshold. Note that the learning process by the learning unit 16 described here is just an example, and the learning unit 16 may learn the upper and lower thresholds using any method.

[0062] Furthermore, the learning unit 16 may learn the rate or amount of increase of the actual feature value per processing operation. For example, the learning unit 16 sets the maximum rate of increase in the transition of the actual feature value where the increasing trend of the actual feature value is normal as the upper limit threshold of the rate of increase, and sets the minimum rate of increase in the transition of the actual feature value where the increasing trend of the actual feature value is normal as the lower limit threshold of the rate of increase.

[0063] In addition, for example, the learning unit 16 sets the maximum increase amount as the upper limit threshold of the increase amount in a transition of the actual feature quantity where the increase trend of the actual feature quantity is normal, and sets the minimum increase amount as the lower limit threshold of the increase amount in a transition of the actual feature quantity where the increase trend of the actual feature quantity is normal.

[0064] The learning unit 16 sets the upper and lower thresholds in a diagnostic model for each machining type, and stores the set diagnostic model in the diagnostic model storage unit 23. The learning unit 16 and the diagnostic model storage unit 23 may be realized on an external device or a CNC device separate from the diagnostic system 1.

[0065] The learning unit 16 may learn the approach section before cutting, the cutting processing section, and the retreat section after cutting based on the waveform of the actual feature amount. In this case, the learning unit 16 sets the average value of the waveforms of multiple actual feature amounts as the reference waveform of the actual feature amount, and sets the approach section before cutting, the cutting processing section, and the retreat section after cutting based on the reference waveform.

[0066] The processing diagnosis unit 17 reads out a diagnosis model corresponding to the processing type from the diagnosis model storage unit 23. The processing diagnosis unit 17 also receives the processing type and the actual feature amount from the corrected feature amount calculation unit 14. The processing diagnosis unit 17 also reads out, from the trend data storage unit 22, the transition of the actual feature amount corresponding to the processing type and the change trend of the feature amount corresponding to the processing type.

[0067] The machining diagnosis unit 17 diagnoses the machining state and the wear state of the tool based on the diagnosis model and the actual feature amount, and diagnoses the machining state based on the change tendency of the feature amount.

[0068] The machining diagnosis unit 17 diagnoses the tool life, machining state (e.g., machining abnormality), etc. based on statistics with the feature values ​​stored in the trend data storage unit 22 as the population and the actual feature values ​​calculated by the corrected feature value calculation unit 14.

[0069] The machining diagnosis unit 17 determines the wear state of the tool by, for example, applying the actual feature amount received from the corrected feature amount calculation unit 14 to the diagnostic model. When the current actual feature amount is equal to or greater than the upper limit threshold of the diagnostic model, the machining diagnosis unit 17 determines that the tool is worn and therefore needs to be replaced.

[0070] Furthermore, when the current actual feature value is equal to or less than the lower limit threshold of the diagnostic model, the machining diagnosis unit 17 determines that the tool is in an abnormal state (for example, the tool is broken) and therefore needs to be replaced. In this way, the upper limit threshold of the diagnostic model is used to diagnose tool wear, and the lower limit threshold of the diagnostic model is used to diagnose the abnormal state of the tool. Furthermore, the upper limit threshold of the diagnostic model may be used to detect a decrease in the cutting amount when the coordinates of the workpiece or the tool are incorrectly corrected.

[0071] Furthermore, the machining diagnosis unit 17 may predict the life of a tool currently in use by applying the transition of the actual feature quantity stored in the trend data storage unit 22 to a diagnostic model. In this case, the diagnostic model calculates a regression line or a curve fit curve, and predicts the life of the tool currently in use based on the trend indicated by the calculated regression line or curve fit curve. The machining diagnosis unit 17 predicts the life of the tool currently in use, for example, by how many more machining operations are required before the actual feature quantity becomes equal to or exceeds the upper limit threshold.

[0072] Furthermore, the machining diagnosis unit 17 may determine the machining state by applying the transition of the actual feature amount stored in the trend data storage unit 22 to the diagnosis model. In this case, the diagnosis model determines that the machining state is abnormal when the change amount, change rate, etc. of the actual feature amount changes suddenly (exceeds a reference value). For example, the diagnosis model determines that the machining state is abnormal when the change amount of the actual feature amount of a tool currently in use becomes equal to or greater than a threshold value set for the change amount of the actual feature amount.

[0073] Furthermore, the machining diagnosis unit 17 diagnoses the machining state based on the change trends of the feature quantities stored in the trend data storage unit 22. For example, the machining diagnosis unit 17 determines that the machining state is abnormal when the change trends of the feature quantities increase sharply or the variations become large. The machining diagnosis unit 17 determines whether the machining state is abnormal based on, for example, a regression line obtained from the change trends of the feature quantities received from the trend data storage unit 22.

[0074] The processing diagnosis unit 17 may estimate the cause of the abnormal processing state based on the change trend (waveform) of the feature amount received from the trend data storage unit 22. In this case, the processing diagnosis unit 17 stores a correspondence relationship between the change trend (waveform) of the feature amount and the cause of the abnormality in the case of this change trend. The processing diagnosis unit 17 identifies the cause of the abnormality based on this correspondence relationship and the change trend of the feature amount.

[0075] For example, the processing diagnosis unit 17 determines that the cause of the abnormality that corresponds to the processing state having the largest correlation coefficient between the change trend of the stored feature quantity and the change trend of the received feature quantity is the cause of the current abnormality.

[0076] Machining diagnosis unit 17 transmits the diagnosis result to diagnosis result output unit 18. If the diagnosis result indicates an abnormal state of machining or a worn state of a tool (replacement time), diagnosis result output unit 18 transmits the diagnosis result to an external device or machine tool 2. This causes machine tool 2 to perform an operation based on the diagnosis result. For example, if diagnosis result output unit 18 transmits information identifying a tool and a diagnosis result indicating that this tool needs to be replaced to machine tool 2, machine tool 2 replaces the tool based on the information identifying the tool.

[0077] Furthermore, if the diagnostic result shows that the transition of the actual feature value or the change trend of the feature value is changing suddenly (for example, a sudden increase), the diagnostic result output unit 18 performs feedback control of the machine tool 2 by outputting an operation control command to the machine tool 2 to ease the machining operation.

[0078] For example, if the actual feature value indicates a value higher than normal, the diagnostic result output unit 18 performs feedback control of the machine tool 2 by outputting an operation control command to the machine tool 2 to slow down the machining operation.

[0079] Furthermore, the diagnostic result output unit 18 may transmit the diagnostic result to a display device (not shown) or the like, and cause the display device to display the diagnostic result.

[0080] Next, the operation of each component in the diagnostic system 1 will be described. Fig. 4 is a flowchart showing the processing procedure of the process executed by the diagnostic system according to the embodiment. After air cutting processing and actual processing are performed in advance for each processing type, the learning unit 16 of the diagnostic system 1 generates a diagnostic model for each processing type based on the actual feature amount. Furthermore, the learning unit 16 of the diagnostic system 1 may generate a diagnostic model for each processing type based on the actual feature amount after G0, G1, G2, or the like in the control command for the processing to be diagnosed is performed.

[0081] Thereafter, when the machine tool 2 starts actual machining, the machining data acquisition unit 10 acquires machining data and machining information from the machine tool 2 (step S10). The machining data acquisition unit 10 identifies the machining type based on the machining information (step S20).

[0082] The section extraction unit 11 extracts sections of the processed data (step S30). Specifically, the correction section extraction unit 12A extracts a correction section of the processed data for each processing type (each diagnostic model). Furthermore, the cutting processing section extraction unit 12B extracts a cutting processing section of the processed data for each processing type.

[0083] The feature amount calculation unit 13 calculates the feature amount of the correction section (step S40). The feature amount calculation unit 13 calculates, for example, the feature amount of the approach section or the tool retraction section as the feature amount of the correction section.

[0084] Furthermore, the feature amount calculation unit 13 calculates the feature amount of the cutting section (step S50). Note that the diagnostic system 1 may execute the process of step S40 and the process of step S50 in any order.

[0085] The corrected feature quantity calculation unit 14 calculates an actual feature quantity based on the feature quantity of the cutting processing section and the feature quantity of the correction section (step S60). When the feature quantity is the average value, maximum value, minimum value, median, or integral value of the processing data, the corrected feature quantity calculation unit 14 calculates the actual feature quantity by subtracting the feature quantity of the correction section from the feature quantity of the cutting processing section. In other words, the corrected feature quantity calculation unit 14 calculates an actual feature quantity from which factors other than cutting processing have been removed by subtracting the feature quantity of the approach section or the retreat section (a feature quantity that does not depend on processing dimensions, etc.) from the feature quantity of the cutting processing section.

[0086] When the feature amount is an integral value of the processed data, the corrected feature amount calculation unit 14 calculates the actual feature amount using the following equation (1).

[0087]

number

[0088] That is, when the feature amount is an integral value of the machining data, the corrected feature amount calculation unit 14 calculates the integral value of the approach section by integrating the spindle load, such as the motor drive current value, of the approach section over the time of the approach section. The corrected feature amount calculation unit 14 also calculates the integral value of the cutting section by integrating the spindle load, such as the motor drive current value, of the cutting section over the time of the cutting section. The corrected feature amount calculation unit 14 multiplies or divides the integral value of the approach section by a value corresponding to the ratio between the length of the approach section and the length of the cutting section. The corrected feature amount calculation unit 14 calculates the actual feature amount, from which factors other than cutting work have been removed, by subtracting the result of the multiplication or division from the integral value of the cutting section.

[0089] In this way, when the feature amount is the integral value of the machining data, the corrected feature amount calculation unit 14 performs subtraction processing according to the ratio between the length of the approach section and the length of the cutting section. In the above-mentioned formula (1), the integral value of the approach section is multiplied by the value obtained by dividing the length of the cutting section by the length of the approach section, but the integral value of the retraction section may also be multiplied by the value obtained by dividing the length of the cutting section by the length of the retraction section. In other words, the corrected feature amount calculation unit 14 may calculate the integral value of the retraction section by integrating the spindle load of the retraction section over the time of the retraction section.

[0090] Furthermore, when the feature value is a range value or standard deviation of the machining data, it is not affected by fluctuations in the spindle load in the approach section and the tool retraction section, so the corrected feature value calculation unit 14 uses the feature value in the cutting processing section as the actual feature value as is.

[0091] The machining diagnosis unit 17 determines the wear state of the tool based on the actual feature amount for each machining type calculated by the corrected feature amount calculation unit 14 (step S70). That is, the machining diagnosis unit 17 performs tool wear diagnosis based on the actual feature amount.

[0092] After the feature amounts of the cutting processing section or the correction processing section are calculated, the change trend calculation section 15 calculates the change trend (amount of change, etc.) of the feature amounts for each processing type (step S65). The processing diagnosis section 17 judges the processing state based on the change trend of the feature amounts for each processing type calculated by the change trend calculation section 15 (step S75).

[0093] In this way, the machining diagnosis unit 17 performs tool wear diagnosis based on the actual feature amount, and also performs machining abnormality diagnosis based on the change trend of the feature amount. Note that the diagnosis system 1 may perform the series of processes of steps S65 and S75 before or after the series of processes of steps S60 and S70. Furthermore, the diagnosis system 1 may perform the series of processes of steps S65 and S75 and the series of processes of steps S60 and S70 in parallel.

[0094] The machining diagnosis unit 17 transmits the diagnosis result to the diagnosis result output unit 18. As a result, the diagnosis result output unit 18 outputs the diagnosis result to the machine tool 2 or an external device such as a display device (step S80). The diagnosis result output unit 18 outputs, for example, an alarm indicating that a tool needs to be replaced. In addition, the diagnosis result output unit 18 outputs, for example, an alarm indicating that the machining state is abnormal.

[0095] For example, when the actual feature value indicates a higher-than-normal value, the diagnostic result output unit 18 performs feedback control of the machine tool 2 by outputting an operation control command to the machine tool 2 to slow down the machining operation. For example, when the workpiece material is expensive or when high-precision machining is required, it may be desirable to perform accurate machining even if it means extending the machining time. In this case, when the feature value increases, the diagnostic system 1 reduces the load on the tool by slowing down the tool feed rate and reducing the amount of cutting per unit time. In this way, the diagnostic system 1 prevents the occurrence of machining defects and can reduce losses due to defective products and the effort required to rework them after they are produced.

[0096] Fig. 5 is a diagram for explaining the process in which the diagnostic system according to the embodiment calculates an actual feature quantity using an integral value of machining data. The horizontal axis of the graph shown in Fig. 5 represents time, and the vertical axis represents the spindle load (such as a motor drive current value). Fig. 5 shows the waveform Wm of the spindle load shown in Fig. 3, the length L2 of the approach section IN2, the length L3 of the actual machining section IN3, and the length L4 of the retreat section IN4.

[0097] One machining process Px includes an approach section IN2, an actual machining section IN3, and an evacuation section IN4. When the feature value is an integral value of the machining data, the feature value differs depending on the length of the section, so the corrected feature value calculation unit 14 performs subtraction processing according to the ratio of the length L2 of the approach section IN2 and the length L3 of the actual machining section IN3 using equation (1) or the like. Note that the corrected feature value calculation unit 14 may use the length L4 of the evacuation section IN4 instead of the length L2 of the approach section IN2.

[0098] 6 is a diagram for explaining the process of determining the machining state and the wear state of the tool based on the feature amount by the diagnostic system according to the embodiment. The horizontal axis of the graph shown in FIG. 6 represents the number of machining operations, and the vertical axis represents the actual feature amount.

[0099] That is, in the graphs in FIG. 6 and subsequent figures, the graphs are generated by plotting the numerical values ​​of the actual feature amounts and the like for each machining operation including the cutting section.

[0100] The diagnostic system 1 calculates the actual feature value for each machining operation. As machining is repeated, the tool wears and the actual feature value increases. Figure 6 shows a graph plotting machining data for each machining operation, for example, the average value of the machining load for each machining operation.

[0101] When the actual feature amount becomes equal to or greater than the upper threshold Uth, the machining diagnosis unit 17 determines that the tool is worn and therefore needs to be replaced. FIG. 6 shows a case where the actual feature amount X1 becomes equal to or greater than the upper threshold Uth at time T1, and the actual feature amount X2 becomes equal to or greater than the upper threshold Uth at time T2. When the machining diagnosis unit 17 determines that a tool replacement is needed, the user replaces the tool with a new tool. That is, after the actual feature amounts X1 and X2 become equal to or greater than the upper threshold Uth, the tool is replaced with a new tool. As a result, the average value of cutting resistance becomes the value when the tool was new.

[0102] The machining diagnosis unit 17 may determine that a tool change is imminent when the difference between the actual feature amount and the upper threshold value Uth is equal to or less than a reference value. In this case, the difference between the actual feature amount and the upper threshold value Uth is displayed on a display device or the like. This allows the user to change the tool to a new tool or continue machining. Note that FIG. 6 shows a case where the tool is changed even when it is determined that a tool change is imminent. FIG. 6 shows a case where the tool is changed twice before time T1.

[0103] Furthermore, the machining diagnosis unit 17 judges the machining state using the actual feature amount and the lower limit threshold Lth. Specifically, when the actual feature amount becomes equal to or less than the lower limit threshold Lth of the diagnosis model, the machining diagnosis unit 17 judges that the tool is in an abnormal state such as broken and therefore needs to be replaced. Fig. 6 shows a case where the actual feature amount X3 becomes equal to or less than the lower limit threshold Lth at the timing of time T3.

[0104] Furthermore, the machining diagnosis unit 17 predicts the life of the tool currently in use based on the transition of the actual feature amount. The machining diagnosis unit 17 predicts the life of the tool (information indicating how many more machining operations are required before the actual feature amount becomes equal to or greater than the upper limit threshold value Uth) based on, for example, the difference between the actual feature amount and the upper limit threshold value Uth and the increase amount or increase rate of the actual feature amount per operation.

[0105] Furthermore, the machining diagnosis unit 17 diagnoses the machining state based on the change tendency of the feature amount stored in the trend data storage unit 22. An example of a method for determining the machining state will be described below.

[0106] FIG. 7 is a diagram for explaining a first processing example in which the diagnostic system according to the embodiment determines the processing state based on the change trend of the actual feature amount. FIG. 8 is a diagram for explaining a second processing example in which the diagnostic system according to the embodiment determines the processing state based on the change trend of the actual feature amount. FIG. 9 is a diagram for explaining a third processing example in which the diagnostic system according to the embodiment determines the processing state based on the change trend of the actual feature amount. FIG. 10 is a diagram for explaining a fourth processing example in which the diagnostic system according to the embodiment determines the processing state based on the change trend of the actual feature amount. FIG. 11 is a diagram for explaining a fifth processing example in which the diagnostic system according to the embodiment determines the processing state based on the change trend of the actual feature amount. FIG. 12 is a diagram for explaining a sixth processing example in which the diagnostic system according to the embodiment determines the processing state based on the change trend of the actual feature amount.

[0107] 7 to 12, the horizontal axis represents the number of times of processing, and the vertical axis represents the actual feature amount. The processing diagnosis unit 17 of the diagnosis system 1 calculates the change trend of the actual feature amount over a plurality of processings (for example, 10 processing sections) from the past processing to the current processing, and determines the processing state based on the calculated change trend of the actual feature amount.

[0108] If there is no change in the machining state, there will be no significant change in the value of the actual feature between previous and next machining cycles or multiple sections, or the actual feature will show a stable upward or downward trend. On the other hand, if there is a change in the machining state as shown below, the machining load will change and a change will appear in the change trend of the actual feature. This will cause a unique change point in the change trend of the actual feature.

[0109] For example, in the machine tool 2, changes in actual feature quantities related to the tool state, changes in actual feature quantities related to the work state, changes in actual feature quantities related to the machining environment, changes in actual feature quantities caused by the work, etc. may occur.

[0110] The graphs shown in Figures 7 to 10 show the changes in the actual feature quantities related to the tool state, and the graph shown in Figure 11 shows the changes in the actual feature quantities related to the workpiece state. Figure 7 shows the changes in the actual feature quantities when the cutting volume decreases at time T4 due to minute chipping on the tool cutting edge.

[0111] Figure 8 shows the change in actual feature quantity when the cutting volume changes due to tool damage caused by the generation or detachment of a built-up edge (BEE) at time T5. A BEE is a part of the cutting edge where melted material from the workpiece adheres to and hardens. A BEE is generated on the tool when melted material from the workpiece adheres to the cutting edge, and when this material separates from the tool, the BEE falls off the tool.

[0112] When a built-up edge is generated and falls off, the cutting volume decreases after the fall-off, and then the cutting volume increases rapidly due to damage to the tool, causing a change in the actual feature value.

[0113] Figure 9 shows the change in the actual feature value when the change in cutting resistance varies greatly for each machining operation due to tool deterioration.

[0114] Figure 10 shows the change in the actual feature value when the change in cutting resistance increases due to the peeling of the tool coating. When the tool coating peels off, the cutting resistance increases sharply.

[0115] FIG. 11 shows the change in the actual feature amount when the cutting resistance changes at time T8 due to a change in the hardness of the workpiece material or the shape of the workpiece material in the previous process or previous machining.

[0116] The pre-processing here refers to the process when the workpiece is formed, such as by casting or forging, and the pre-processing refers to the processing before the workpiece is cut, such as rough machining. For example, when the material lot of the workpiece is changed, the hardness or shape of the workpiece material changes, and the cutting resistance suddenly changes, causing a change in the actual feature value.

[0117] 12 shows a graph of a case where cutting resistance suddenly increases at time T9 due to a change in the machining environment. For example, if the coolant discharge pressure or discharge rate decreases, if the coolant discharge stops, or if the coolant discharge direction changes, the actual feature value changes due to a sudden increase in cutting resistance.

[0118] In addition, if a change due to the work occurs, the cutting volume or cutting resistance will also change. For example, if the work of tool compensation (correction of coordinates according to the length or diameter of the tool) is inappropriate, the cutting volume will rise sharply due to the change due to the work. In this case, the graph will be similar to the graph in Figure 12.

[0119] As shown in Figures 7 to 12, even when the machining state is abnormal, the machining state often does not change to a level that greatly exceeds the upper limit threshold Uth, etc. For example, as shown in Figure 7, when a small tool chipping occurs, the actual feature amount (cutting volume) decreases, causing a relative decrease in the spindle load due to cutting, but the change in the actual feature amount is small. Therefore, in order to detect small changes in the actual feature amount, the machining diagnosis unit 17 may calculate the standard deviation of the actual feature amount in a plurality of machining sections in which machining was performed under the same machining conditions, and judge the machining state based on the standard deviation of changes from the past.

[0120] 13 is a diagram for explaining a first processing example in which the diagnostic system according to the embodiment determines the processing state based on the actual feature amount. In the first processing example, the diagnostic system 1 determines the processing state based on the amount of change (standard deviation) of the actual feature amount.

[0121] Graph GR1 on the left side of the graphs shown in FIG. 13 is the graph shown in FIG. 7, and shows the transition of the actual feature amount. Graph GR2 in the center of the graphs shown in FIG. 13 is a graph showing, as a machining section D1, a plurality of machining sections (for example, 10 machining sections) that are targets for abnormality judgment of the machining state in graph GR1. Graph GR3 on the right side of the graphs shown in FIG. 13 is a graph showing the amount of change (standard deviation) of the actual feature amount of graph GR2. The horizontal axis of graph GR3 shown in FIG. 13 is the number of machining operations, and the vertical axis is the standard deviation. Graph GR3 shows the standard deviation for a certain number of machining operations, including past times each time machining has been performed, for the entire machining section.

[0122] The processing diagnosis unit 17 calculates the standard deviation of the actual feature values ​​to calculate the amount of change in the actual feature values ​​for each of the multiple processing sections. As a result, a change trend in the standard deviation is calculated as shown in graph GR3 with respect to the transition of the feature values ​​in graph GR2. The processing diagnosis unit 17 highlights small changes in the actual feature values ​​by calculating the standard deviation of a specific number of processings (e.g., 10 processings) from the previous processing to the current processing for each processing. By determining the processing state based on the standard deviation of the actual feature values ​​in multiple processing sections D1, the processing diagnosis unit 17 can determine the processing state while excluding sudden changes in the actual feature values ​​(such as false detection).

[0123] When the processing state is normal, the amount of change in the actual feature amount is stable. In other words, when the processing state is normal, the difference (amount of change) between the actual feature amount one time before and the actual feature amount of this time is stable and does not fluctuate greatly with each processing.

[0124] For example, when the processing diagnosis unit 17 calculates the standard deviation for the processing section D1, as shown in graph GR3, the standard deviation St for the processing section D1 stands out more than the standard deviations for the other sections. That is, the amount of change in the standard deviation is greater when the processing state is abnormal than the change in the actual feature amount. For example, when the currently completed processing in the processing section D1 is the Xth (X is a natural number) processing, the processing diagnosis unit 17 calculates the standard deviation of the actual feature amount from the most recent Xth processing to the (X-9)th processing. When the (X+1)th processing is completed, the processing diagnosis unit 17 calculates the standard deviation of the actual feature amount from the (X+1)th processing to the (X-8)th processing.

[0125] When the calculated standard deviation of the machining section D1 is equal to or greater than the standard deviation threshold Th, the machining diagnosis section 17 determines that the machining state of the machining section D1 for which the standard deviation was calculated is abnormal. The machining diagnosis section 17 sets the threshold Th based on, for example, a change trend of the standard deviation, and detects an abnormal state based on the threshold Th. The threshold Th may be set by the user.

[0126] In this way, by determining the processing state based on the standard deviation of the actual feature amounts in a plurality of processing sections in which processing was performed under the same processing conditions, the processing diagnosis unit 17 can determine the processing state based on a more prominent change than when the processing state is determined based on the actual feature amounts as described with reference to Fig. 7. Therefore, by determining the processing state based on the standard deviation of the actual feature amounts in a plurality of processing sections in which processing was performed under the same processing conditions, the processing diagnosis unit 17 can determine the processing state more easily and accurately.

[0127] Next, a second processing example will be described in which the diagnostic system 1 determines the processing state based on the actual feature amount. In the second processing example, the diagnostic system 1 determines the processing state based on the variation of the actual feature amount from the regression line.

[0128] The processing diagnosis unit 17 calculates a regression line (regression equation) of the actual feature values ​​for multiple processing sections (e.g., 10 processing sections) from past processing to current processing, and calculates the variation (standard deviation) from the calculated regression line using the following equation (2).

[0129]

number

[0130] In formula (2), St is the variation (standard deviation) from the regression line, and n is the number of processing sections for which the regression line is calculated. i is each actual feature value from the latest to the nth (n is a natural number) cutting section, which is the section to be diagnosed, X is the value of the corresponding point calculated from the regression equation (x = aN + b), and N is the number of times each tool is used from the latest to the nth cutting section. Note that a (regression coefficient) and b (intercept) of the regression equation are values ​​calculated in the section to be diagnosed.

[0131] 14 is a diagram for explaining a second processing example in which the diagnostic system according to the embodiment determines the processing state based on the actual feature amount. The horizontal axis of the graph shown in FIG. 14 represents the number of processing times, and the vertical axis represents the actual feature amount.

[0132] 14 shows the transition of the actual feature amount and a regression line Y1 of the actual feature amount in the processing section D2 that is the processing state judgment target. When the actual feature amount shows a stable upward trend, the deviation from the regression line Y1 is small, but when the deviation from the regression line Y1 is large, the standard deviation from the regression line Y1 becomes large.

[0133] The machining diagnosis unit 17 calculates a regression line Y1 from the actual feature amounts in the machining section D2, which is the diagnosis target section and is counted back to the most recent cutting machining section up to the nth (n is a natural number) cutting machining section, and calculates the residual of the actual feature amount from the regression line Y1 for the machining section D2. FIG. 14 shows a case where the residual amount for each machining operation in the machining section D2 is calculated. The machining diagnosis unit 17 determines whether the machining state is abnormal or normal based on whether the square root of the residual of the actual feature amount from the regression line Y1 is equal to or greater than a threshold value. In this way, the machining diagnosis unit 17 can grasp the change trend of the actual feature amount by calculating the standard deviation from the regression line Y1.

[0134] 15 is a diagram for explaining a third processing example in which the diagnostic system according to the embodiment determines the processing state based on the actual feature amount. In the third processing example, the diagnostic system 1 determines the processing state based on a regression line Y1 of the actual feature amount. The horizontal axis of the graph shown in FIG. 15 represents the number of processing times, and the vertical axis represents the actual feature amount.

[0135] In the third processing example, the processing diagnosis unit 17 calculates the regression line Y1 by the same method as in the second processing example. The processing diagnosis unit 17 determines the processing state based on at least one of the change tendency of the slope and the change tendency of the intercept of the calculated regression line Y1.

[0136] The machining diagnosis unit 17 determines that the machining state is abnormal, for example, when the slope of the regression line Y1 exceeds a reference range. Also, the machining diagnosis unit 17 determines that the machining state is abnormal, for example, when the intercept of the regression line Y1 exceeds a reference range.

[0137] The reference range of the slope may be set by the processing and diagnosing unit 17 based on past slopes, or may be set by the user. The processing and diagnosing unit 17 sets the reference range of the slope based on, for example, the average value, standard deviation, etc. of past slopes.

[0138] The reference range of the intercept may be calculated by the processing and diagnosis unit 17 based on past intercepts, or may be set by the user. The processing and diagnosis unit 17 sets the reference range of the intercept based on, for example, the average value, standard deviation, etc. of past intercepts.

[0139] When a tool deteriorates suddenly, the condition of the cutting edge of the tool changes significantly with each machining operation, resulting in large fluctuations in cutting resistance. Fig. 16 is a diagram for explaining an example of the transition of actual feature values ​​calculated by the diagnostic system according to the embodiment when a tool deteriorates. The horizontal axis of the graph shown in Fig. 16 represents the number of machining operations, and the vertical axis represents actual feature values.

[0140] In the diagnostic system 1, depending on the type of machining, the cutting resistance fluctuates slightly with each machining. In this case, if the tool deteriorates suddenly, the amount of change in cutting resistance will fluctuate more than before the deterioration. Furthermore, if the tool deteriorates suddenly, the cutting resistance may change from an increasing trend to a decreasing trend. FIG. 16 shows a case where the actual feature value changes from an increasing trend to a decreasing trend due to the sudden deterioration of the tool at time T10, and the actual feature value repeatedly fluctuates greatly up and down with each machining.

[0141] Fig. 17 is a diagram for explaining a process in which the diagnostic system according to the embodiment determines the machining state in a second processing example for the graph shown in Fig. 16. In the second processing example, the diagnostic system 1 determines the machining state based on the change trend of the variation (standard deviation) of the actual feature value from the regression line. The horizontal axis of the graph shown in Fig. 17 represents the number of machining operations, and the vertical axis represents the standard deviation of the actual feature value from the regression line. Fig. 17 shows the change trend of the standard deviation of the actual feature value from the regression line.

[0142] The machining diagnosis unit 17 calculates a regression line for the actual feature amounts shown in Fig. 16 and calculates the amount of deviation (standard deviation) of the actual feature amounts from the regression line. Then, the machining diagnosis unit 17 determines whether the machining state is abnormal or normal based on whether the standard deviation of the actual feature amounts from the regression line is equal to or greater than a threshold value Th. Fig. 17 shows a case where the standard deviation of the actual feature amounts from the regression line increases due to tool deterioration at time T10.

[0143] FIG. 18 is a diagram for explaining a first example of a process in which the diagnostic system according to the embodiment determines the machining state in the third processing example for the graph shown in FIG. 16. In the first example of the third processing example, the diagnostic system 1 determines the machining state based on the change trend of the slope of the regression line of the actual feature amount. The horizontal axis of the graph shown in FIG. 18 represents the number of machining times, and the vertical axis represents the slope of the regression line. FIG. 18 shows the change trend of the slope of the regression line.

[0144] The machining diagnosis unit 17 calculates a regression line for the actual feature values ​​shown in Fig. 16 and calculates the slope of the calculated regression line. For example, the machining diagnosis unit 17 determines that the machining state is abnormal when the slope of the regression line exceeds a reference range. Fig. 18 shows a case where the slope of the regression line decreases due to tool deterioration and becomes equal to or smaller than the lower threshold value Th1 of the slope of the regression line at time T10.

[0145] FIG. 19 is a diagram for explaining a second example of processing in which the diagnostic system according to the embodiment determines the processing state in the third processing example for the graph shown in FIG. 16. In the second example of the third processing example, the diagnostic system 1 determines the processing state based on the change trend of the intercept of the regression line of the actual feature amount. The horizontal axis of the graph shown in FIG. 19 is the number of processing times, and the vertical axis is the intercept of the regression line. FIG. 19 shows the change trend of the intercept of the regression line.

[0146] The machining diagnosis unit 17 calculates a regression line for the actual feature amounts shown in Fig. 16 and calculates an intercept of the calculated regression line. For example, the machining diagnosis unit 17 determines that the machining state is abnormal when the intercept of the regression line exceeds a reference range. Fig. 19 shows a case where the intercept of the regression line increases due to tool deterioration and becomes equal to or exceeds the upper threshold value Th2 of the intercept of the regression line at time T10. Fig. 19 also shows a case where the intercept of the regression line increases due to tool deterioration at time T10.

[0147] In the graph shown in FIG. 16, it is difficult to detect abnormalities in the change trend of the actual feature amount, but the diagnostic system 1 can easily determine the processing state and easily detect abnormalities by using any of the methods described in FIGS. 17 to 19.

[0148] FIG. 20 is a diagram for explaining spindle load data used by the diagnostic system of the comparative example. The horizontal axis of the graph shown in FIG. 20 represents time, and the vertical axis represents spindle load. The diagnostic system of the comparative example calculates feature values ​​based on the spindle load (motor drive current value, etc.) that includes influences other than fluctuations due to actual machining, and diagnoses the wear state of the tool based on these feature values. In the diagnostic system of the comparative example, physical quantities such as the motor drive current value may change due to temperature changes, deterioration state, etc. of the machine tool 2. The following three factors can be cited as factors that cause physical quantities to change. (Factor F1) Increase or decrease in operating resistance (such as bearing friction resistance) or sliding resistance due to temperature changes in machine tool 2 or the motor (Factor F2) Increase or decrease in operating resistance or sliding resistance due to maintenance or part replacement of machine tool 2 (Factor F3) Increase or decrease in rotational resistance or rotational force due to the length of the bar material in an automatic lathe

[0149] Sliding resistance is the dynamic friction that occurs when, for example, a ball screw or LM (Linear Motion) guide rotates or moves back and forth. Bar stock is a long, round workpiece, and after one cutting process, the machined portion is cut off and the next cutting process begins. In the case of bar stock, as the machined portion is cut off, the length becomes shorter as the number of times it is cut increases, and the rotational resistance or rotational force during processing changes.

[0150] FIG. 20 shows a waveform W1 of the spindle load during normal operation and a waveform W2 when the spindle load increases due to factors other than machining. For example, when the machine tool 2 has not been warmed up and starts machining in a cold state, as shown in FIG. 20, machining data such as the spindle load may increase overall, resulting in waveform W2. In this case, the diagnostic system of the comparative example calculates feature quantities based on the increased machining data. Therefore, the diagnostic system of the comparative example calculates unique feature quantities that are different from those during normal continuous operation, and determines the wear state of the tool, etc., based on the unique feature quantities.

[0151] In this way, when the processing data fluctuates due to factors other than processing, the diagnostic system of the comparative example determines the wear state of the tool based on the feature that expresses the increase or decrease, and therefore may not be able to accurately determine the wear state of the tool.

[0152] FIG. 21 is a diagram for explaining the feature quantities used by the diagnostic system of the comparative example. The horizontal axis of the graph shown in FIG. 21 represents the number of machining operations, and the vertical axis represents the actual feature quantities. When tool breakage or a significant machining abnormality occurs, the feature quantities change significantly and exceed the upper threshold value Uth or the lower threshold value Lth, so that even the diagnostic system of the comparative example can detect an abnormality in the machining state. However, in the case of a machining abnormality in which the tool is slightly damaged or the machining state changes only slightly, as shown in data d1 to d3 of FIG. 21, the feature quantities do not exceed the upper threshold value Uth or the lower threshold value Lth, and the diagnostic system of the comparative example could not detect an abnormality in the machining state.

[0153] On the other hand, the diagnostic system 1 judges the machining state based on the change tendency of the feature amount, and therefore can detect tool loss or machining abnormalities in which the machining state changes only slightly.

[0154] Fig. 22 is a diagram for explaining the timing of tool replacement when the diagnostic system according to the embodiment determines the wear state of the tool. Fig. 22 shows the timing of tool replacement when the diagnostic system 1 determines the wear state of the tool and the timing of tool replacement when the diagnostic system of the comparative example determines the wear state of the tool. The horizontal axis of Fig. 22 represents the number of machining operations, and the vertical axis represents the actual feature amount.

[0155] Since the diagnostic system of the comparative example calculates the feature quantity based on the motor drive current value and the like that include influences other than those caused by the fluctuations due to the actual machining, the feature quantity may fluctuate due to factors unrelated to the actual machining, such as sliding resistance, as described above. In this case, even if the machining state is normal, the diagnostic system of the comparative example may diagnose a sudden increase or decrease in the feature quantity as an abnormality.

[0156] The diagnostic system 1 of the embodiment automatically finds the actual machining section IN3, extracts only the amount of variation in the feature due to the actual machining, and calculates the actual feature by removing variations unrelated to the machining. Therefore, it can be said that the amount of variation in the actual feature calculated by the diagnostic system 1 of the embodiment is an accurate amount of variation due to the actual machining. As a result, the diagnostic system 1 of the embodiment can capture minute changes due to the actual machining even when the change in the actual feature calculated in cutting is small, and can easily detect abnormalities in the wear state of the tool or the machining state.

[0157] The diagnostic system of the comparative example determines the timing of tool replacement based on the number of times of machining (50 times in Figure 22), for example, so there are cases where the tool is replaced even though it has not actually reached the end of its life, and cases where the tool is not replaced even though its life has expired.

[0158] On the other hand, the diagnostic system 1 of the embodiment can accurately determine the wear state of a tool based on the actual feature amount. This allows the user to replace the tool at an appropriate timing depending on the wear state of the tool. For example, if the actual feature amount increases slowly due to tool wear, the diagnostic system 1 of the embodiment can perform more machining than the number of machining cycles (50) specified by the diagnostic system of the comparative example, thereby reducing the cost associated with tool replacement. Figure 22 shows a case where the diagnostic system 1 of the embodiment replaced the tool after 70 machining cycles when the progress of tool wear was slow (C1).

[0159] Furthermore, in the diagnostic system of the comparative example, when tool deterioration or wear progressed rapidly, the tool was not replaced, resulting in a machining defect. On the other hand, the diagnostic system 1 of the embodiment can replace the tool regardless of the number of machining operations when tool deterioration or wear progresses rapidly, thereby preventing the occurrence of machining defects caused by deterioration or wear. Furthermore, the diagnostic system 1 of the embodiment can detect the occurrence of sudden machining defects, thereby preventing the outflow of defective products. Figure 22 shows a case where the diagnostic system 1 of the embodiment replaced the tool after 35 machining operations when tool deterioration or wear progressed rapidly (C2).

[0160] In this way, by removing elements unrelated to the actual cutting load from the machining data, the diagnostic system 1 can reduce the possibility of diagnosing and detecting machining when it is obscured by factors other than machining (e.g., sliding friction resistance), and can also reduce erroneous diagnosis and detection due to factors other than machining.

[0161] As a result, the diagnostic system 1 can stabilize process quality by diagnosing all workpieces, preventing the occurrence and outflow of defective products. Therefore, the diagnostic system 1 can eliminate processes such as visual inspection, sampling inspection, and image inspection, thereby reducing the number of man-hours.

[0162] The diagnostic system 1 may be realized by a single computer, or may be realized by multiple computers or CNC devices connected via a network. Here, the hardware configuration of the diagnostic system 1 when the diagnostic system 1 is realized by a single computer will be described.

[0163] The diagnostic system 1 is realized by a processing circuit, which may be a processor and memory that executes a program stored in memory, or may be dedicated hardware.

[0164] FIG. 23 is a diagram illustrating an example of the configuration of a processing circuit included in a diagnostic system according to an embodiment, where the processing circuit is realized by a processor and a memory. The processing circuit 90 illustrated in FIG. 23 includes a processor 91 and a memory 92. When the processing circuit 90 includes the processor 91 and the memory 92, each function of the processing circuit 90 is realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a diagnostic program and stored in the memory 92. In the processing circuit 90, each function is realized by the processor 91 reading and executing the diagnostic program stored in the memory 92. That is, the processing circuit 90 includes the memory 92 for storing the diagnostic program that results in the processing of the diagnostic system 1. This diagnostic program can also be said to be a program that causes the diagnostic system 1 to execute each function realized by the processing circuit 90. This diagnostic program may be provided by a computer-readable recording medium on which the diagnostic program is recorded, or by other means such as a communication medium.

[0165] The diagnostic program can also be said to be a program that causes the diagnostic system 1 to execute the processes of steps S10 to S80 in Fig. 4. Here, the processor 91 is, for example, a CPU (Central Processing Unit), a processing unit, an arithmetic unit, a microprocessor, a microcomputer, or a DSP (Digital Signal Processor). Also, the memory 92 is, for example, a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable ROM), or an EEPROM (Electrically EPROM), a magnetic disk, a flexible disk, an optical disk, a compact disk, a minidisk, or a DVD (Digital Versatile Disc).

[0166] FIG. 24 is a diagram illustrating an example of the configuration of a processing circuit included in a diagnostic system according to an embodiment, when the processing circuit is configured with dedicated hardware. The processing circuit 93 illustrated in FIG. 24 corresponds to, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof. The processing circuit 93 may be partially implemented with dedicated hardware and partially implemented with software or firmware. In this way, the processing circuit 93 can realize each of the above-described functions by dedicated hardware, software, firmware, or a combination thereof.

[0167] As described above, the diagnostic system 1 of the embodiment calculates actual feature values ​​corresponding to the actual cutting load, which is the load during cutting, by subtracting the feature values ​​in the correction section from the feature values ​​in the cutting section, and performs machining diagnosis based on the actual feature values. This allows the diagnostic system 1 to perform machining diagnosis based on the actual feature values ​​corresponding to the actual cutting load during cutting, with factors other than machining, such as cold conditions, removed. Therefore, the diagnostic system 1 can achieve highly accurate machining diagnosis even when the change in the feature values ​​of the machining data calculated from the machining data during cutting is small.

[0168] Furthermore, the diagnostic system 1 extracts the cutting processing section and the correction section based on processing data obtained when air cutting processing and actual processing are performed in advance, so that the cutting processing section and the correction section can be extracted accurately.

[0169] Furthermore, the diagnostic system 1 calculates the change trend of the feature amount of the machining data from the past to the present, and performs machining diagnosis based on the calculated change trend. As a result, the diagnostic system 1 can realize highly accurate machining diagnosis even when the change in the feature amount of the machining data calculated from the machining data in cutting is small.

[0170] Furthermore, the diagnostic system 1 determines the machining state of the machine tool 2 based on the standard deviation of the change trend, so that highly accurate machining diagnosis can be achieved even when the change in the feature amount of the machining data is even smaller.

[0171] Furthermore, the diagnostic system 1 determines the machining state of the machine tool 2 based on the regression line of the change trend, so that highly accurate machining diagnosis can be achieved even when the change in the feature amount of the machining data is even smaller.

[0172] The configurations shown in the above embodiments are merely examples, and may be combined with other known technologies, and parts of the configurations may be omitted or modified without departing from the spirit of the invention. [Explanation of symbols]

[0173] 1 diagnostic system, 2 machine tool, 10 machining data acquisition unit, 11 section extraction unit, 12A correction section extraction unit, 12B cutting processing section extraction unit, 13 feature calculation unit, 14 corrected feature calculation unit, 15 change trend calculation unit, 16 learning unit, 17 machining diagnosis unit, 18 diagnosis result output unit, 22 trend data storage unit, 23 diagnosis model storage unit, 90, 93 processing circuit, 91 processor, 92 memory, d1 to d3 data, D1, D2 machining section, F1 to F3 factors, GR1 to GR3 graph, IN1 acceleration section, IN2 approach section, IN3 actual machining section, IN4 withdrawal section, IN5 deceleration period, L2 to L4 length, Lb base load, Lc actual cutting load, Lth lower limit threshold, Rm, W1, W2, Wm waveform, St Standard deviation, T1, T2, T3, T4, T5, T8, T9, T10 time, Th threshold, Uth upper threshold, X1~X3 actual feature, Y1 regression line.

Claims

1. a machining data acquisition unit that acquires machining data from the start to the end of machining by the machine tool; a section extraction unit that extracts a first waveform of a first section, which is a cutting section, and a second waveform of a second section, which is a section other than the first section, from the waveform of the acquired machining data based on a waveform of pre-machining data, which is machining data when air cutting machining and actual machining are previously performed; a feature amount calculation unit that calculates a first feature amount in the first section of the first waveform and a second feature amount in the second section of the second waveform; a corrected feature amount calculation unit that calculates a third feature amount corresponding to an actual cutting load, which is a load of cutting work, by subtracting the second feature amount from the first feature amount; a processing diagnosis unit that executes processing diagnosis based on the third feature amount; Equipped with A diagnostic system characterized by:

2. further comprising a learning unit that learns the first section and the second section based on the processed data; The diagnostic system of claim 1 .

3. the learning unit learns, based on the past third feature amount, a first threshold value of the third feature amount for determining whether or not a tool used by the machine tool needs to be replaced, and a second threshold value of the third feature amount for determining whether or not the tool is in an abnormal state; the machining diagnosis unit determines whether or not to replace the tool based on the first threshold value and the current third feature amount, and determines whether or not the tool is in an abnormal state based on the second threshold value and the current third feature amount. The diagnostic system according to claim 2 .

4. further comprising a diagnosis result output unit that outputs a diagnosis result of the machining diagnosis to an external device or the machine tool; 4. The diagnostic system according to claim 1, wherein the diagnostic system comprises: a first detecting means for detecting a first error;

5. the diagnostic result output unit outputs an operation control command corresponding to the diagnostic result to the machine tool, thereby performing feedback control of the machine tool. The diagnostic system according to claim 4 .

6. a machining data acquisition step in which a diagnostic system for diagnosing machining by a machine tool acquires machining data from the start to the end of machining by the machine tool; a section extraction step in which the diagnostic system extracts a first waveform of a first section, which is a cutting section, and a second waveform of a second section, which is a section other than the first section, from the waveform of the acquired machining data based on a waveform of pre-machining data, which is machining data when air cutting machining and actual machining are previously performed; a feature amount calculation step in which the diagnostic system calculates a first feature amount in the first section of the first waveform and a second feature amount in the second section of the second waveform; a corrected feature quantity calculation step in which the diagnostic system calculates a third feature quantity corresponding to an actual cutting load, which is a load of cutting work, by subtracting the second feature quantity from the first feature quantity; a process diagnosis step in which the diagnosis system executes a process diagnosis based on the third feature amount; Including, A diagnostic method characterized by:

7. a machining data acquisition step of acquiring machining data from the start of machining by the machine tool to the end of machining; a section extraction step of extracting a first waveform of a first section which is a cutting section and a second waveform of a second section which is a section other than the first section from the waveform of the acquired machining data based on a waveform of pre-machining data which is machining data when air cutting machining and actual machining are previously performed; a feature calculation step of calculating a first feature in the first section of the first waveform and a second feature in the second section of the second waveform; a corrected feature value calculation step of calculating a third feature value corresponding to an actual cutting load, which is a load of cutting work, by subtracting the second feature value from the first feature value; a machining diagnosis step of performing machining diagnosis based on the third feature amount; to the computer, A diagnostic program characterized by:

Images