JPS6144623B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a cutting abnormality detection device, and particularly to a cutting abnormality detection device in milling.

Recently, with the spread of NC machine tools, unmanned operation of machine tools and having multiple machines have become commonplace. However, the NC machines that are currently becoming popular have almost no function to detect when a tool breaks, so machining continues, resulting in serious damage to the workpiece and the tool.

For these reasons, interest in detecting tool breakage is increasing. Recently, the importance of this problem has been pointed out, and various detection devices for detecting breakage and wear of drills have been proposed.

Another technical direction belongs to the field of adaptive control, which has already been put into practical use in some areas. In this method, for example, physical quantities (depth of cut, feed, etc.) between the workpiece and the tool are controlled so that the spindle torque during cutting does not exceed a preset value. However, in this method, the constant command value itself from the NC device is actually made into a variable amount, and it is also connected to the feed drive system of the workpiece such as a table, making the control system complex and causing wear. It is not perfect in that the cutting performance of the tool deteriorates as it progresses, and the phenomenon of breakage is unavoidable.

On the other hand, detection of cutting abnormalities, mainly breakage, in milling, which is the second most commonly used cutting tool after drilling, has not been published so far, and the technical methodology is quite different from that of drilling, so it is an undeveloped area. It had become a field. Furthermore, as mentioned above, there are examples of using milling cutters for adaptive control of machine tools, but we have not yet reached the stage where new technical ideas are presented by linking this to breakage and other problems.

As mentioned previously, in the case of milling, parameters such as depth of cut, feed rate, outside diameter of the milling cutter, how many chips each milling cutter consists of, and cutting width are compared to drills. There are many parameters, and milling cutters are relatively strong tools compared to drills, so even if you try to conduct an experiment with several parameters, the experiment itself will require a considerable amount of work, and the differences between the obtained data will vary. One example of this is the difficulty in evaluating mutual relationships, which exists as a technical barrier for engineers.

The present inventors solved this technical problem by repeating various experiments as described below, and at the stage of organizing the obtained data, we investigated the cutting width (W) of the spindle current value associated with the intermittent cutting phenomenon unique to milling. ) and its average value (R) during cutting. In other words, dimensionless quantity
We have discovered a relationship in which W/R (or R/W) can be distinguished between a broken state and a normal state without depending on most of the parameters in milling mentioned above. By utilizing this relationship, he invented a device that detects abnormalities such as breakage during milling.

In the following explanation, we will first explain the main points of the above-mentioned experimental process, and then explain the dimensionless quantity (W/
We will discuss the significance of paying attention to R).

Experimental process The machine tool used in the experiment was a gate-type horizontal boring machine with a 6-inch (10-blade) milling tool.
and 5 inches (8 blades) were used.

Fig. 1 is a block diagram of the experimental data measurement section, from the shunt 11 that detects the shunt current of the main shaft drive motor (DC motor) drive unit to which the milling cutter is mounted, to the DC amplifiers 12, 12.
The signal is amplified by A and stored on magnetic tape by a data recorder. DC (for later frequency analysis)
The output of amplifier 12A is low pass filter LPF14
The data is recorded on recording paper by the pen recorder 15 via the pen recorder 15. Vibration-related signals are generated by piezoelectric pickup 16.
The signal is detected and recorded on another channel of the data recorder 13 through a charge amplifier 17 and a filter 18. Also, the vibration signal passing through the filter 18 is
It is adapted to be recorded by a pen recorder 20 via an RMS (root mean square) converter 19.

Figures 2 and 3 are diagrams showing the locus of the cutter center when cutting with the 6-inch and 5-inch milling cutters, and in Figure 2 A and B, the width is 100 mm.
A workpiece 51 with a length of 450 mm and a length of 450 mm is cut along a path such that the center of the cutter 52 is →→→.
Figure A is a plan view, B is a front view, and the solid line indicates cutting feed.The broken line indicates rapid feed. In Figure 3 A and B, the cutting width is
Cutting was performed using the cutter 53 along a path such as →→→→→→ so that two types of machining, 83 mm and 17 mm, could be performed.

In Fig. 4, a workpiece 51 to be cut is placed on a table 6.
1 shows the installed and fixed state. 51A is the surface to be cut.

Figure 5 shows four types of missing blades created by scratching a part of the milling blade with an electric pen and changing its position and shape. One of the sections is adapted to be set into a cutting position.

In contrast to the above-mentioned numbers No. 3 to No. 6, as will be referred to later, No. 1 is cut by a normal blade. or
No. 2 corresponds to a single chipped chip, and is the case when only one chip is removed from the milling cutter.

Above, No. 2 to No. 6 are (a) The blade is chipped and no longer participates in cutting (No. 2) (b) The blade is scratched and cutting quality is reduced (No. 3 to No. 6)
No.6) (c) The cutting edge has deteriorated due to wear (No.3~
No. 6).

6 to 8 are waveform diagrams of main shaft motor current values detected through a 10 Hz low-pass filter. Other parameters are as follows.

Milling cutter used: 6 inches (10 blades) Cutting speed: 90 m/min Spindle speed: S = 190 rpm Feed rate: 0.25 mm/tooth (Feed rate: F = 475 mm/min) Depth of cut: h = 1, 2, 3 (m) Cutting width 100mm Under the above conditions, Fig. 6a shows the current waveforms for each case of a normal blade (No. 1) and cutting depth h = 1, 2, and 3mm. Figure b shows the case of one chipped edge (No. 2), and compared to figure a, the variation width value W tends to increase. Fig. 7c shows the case of a defective blade (No. 3), and Fig. 7d shows the case of a defective blade (No. 4). Further, Fig. 8e shows a defective blade (No. 5), and Fig. 8f shows a defective blade (No. 6). Observing Figures c, d, e, and f, it can be seen that the fluctuation width W is smaller in all of them than in Figure b. On the other hand, the average value R varies depending on the depth of cut h, but if h is the same, there is no difference in a to f. Also, as shown in Figure b, the lowest value (VB) just before the peak (VP) during one rotation corresponds to one tooth being missing, and the normal depth of cut is 0.25 mm at the peak. /
This is thought to be because it gives twice as much power as the blade.

FIGS. 9a and 9b are waveform diagrams corresponding to FIGS. 6a and 6b, respectively, and the difference is that the waveforms were recorded through a 50Hz low-pass filter instead of 10Hz.
Therefore, it can be seen that a frequency component between 10 Hz and 50 Hz exists (although the pen recorder does not respond up to several tens of Hz).

In addition, the condition of the missing blades of No. 3 to No. 6 mentioned above is shown in Figure b.
Compared to (No. 2), which has one chipped edge, it is considered to be closer to the normal blade (No. 1), but even if the chipped edge is not completely chipped, when the chipped edge no longer participates in cutting, It should be noted that this is a blade.

FIGS. 10 and 11 show the results of frequency analysis of the waveforms shown in FIGS. 6a and 6b, respectively, using a spectrum analyzer. In each figure, the horizontal axis shows the frequency (Hz), and the vertical axis shows the power spectrum of the current. Looking at this result, the frequency components of the current can be roughly divided into two parts (bands).

One is the lower frequency component, and the main components are 3.2Hz and 6.3Hz. The other component is a relatively high frequency component distributed in a band of about 30Hz to 40Hz. Here, the 3.2Hz component is due to the rotation of the main axis.

That is, since the spindle rotation speed is 190 rpm, 190/60≒3.2 (Hz), and it will be understood that 3.2 Hz is the frequency corresponding to the spindle rotation speed (hereinafter referred to as the fundamental frequency).

Also, 6.3Hz is twice that frequency component. On the other hand, regarding the higher frequency component, since this experiment used a 6-inch, 10-blade milling cutter, this band includes the fundamental frequency times the number of blades, or 32 Hz. Other components are machine-specific components. (From the analysis results of no-load current, not shown) 10th
Comparing Figures 1 and 11, when cutting with a normal blade (Figure 10), each component increases uniformly as the depth of cut (h) increases, but when cutting with one chipped edge (Figure 11) In Figure), it can be seen that the increase in low frequency components is significant. As mentioned above, when one blade is removed, cutting is not performed even when the blade should start cutting, so the cutting current value becomes temporarily small, and then the next blade starts cutting. This can be explained as a phenomenon in which the current value rapidly increases because when cutting, it is necessary to cut by the amount of feed for two blades, including that of the immediately preceding blade. Therefore, since such fluctuations in the cutting current value appear every rotation of the spindle, the frequency component of 3.2 Hz, which corresponds to the spindle rotation speed, increases. Furthermore, since the current value fluctuates rapidly, high frequency components of 3.2 Hz, ie, components of 6.3 Hz, 9.5 Hz, etc. also increase.

The above experiment shows that when a breakage occurs, there is a significant change in the low frequency components.On the contrary, the higher frequency components do not change much, so if you cut out the higher frequency components with a low pass filter, there will be a noticeable difference in the breakage. can be expected to appear.

Figures 12 and 13 show the power spectrum of a 5-inch (8-blade) milling cutter with one cutting edge after passing through a 10Hz Ropa filter. Figure 13 shows the case where the cutting width is 83 mm.

The power spectrum for the normal blade corresponding to Figures 12 and 13 is not shown, but from the same figure it is 3.8Hz,
It can be easily inferred that components such as 6.8Hz and 6.6Hz are larger than the frequency components.

Figure 14 is an example of data for observing changes in power spectrum when the spindle rotation speed is changed, and the spindle rotation speed S = 226 rpm (cutting speed 90 m/min).
Cutting was carried out at four different rotation speeds: 100%, 90%, 80%, and 70%. Other cutting conditions are as follows.

Milling cutter used: 5 inches (8 blades) Feed rate: 0.25 mm/Blade depth of cut: 1, 2, 3 mm (However, only 3 mm is shown in the figure) Cutting width: 83 mm, 17 mm What can be determined from Figure 14 is that the spindle speed 30.0Hz, 27.0Hz, 23.9Hz, 21.1Hz, etc. are shown in response to the change.

Figures 15 and 16 are power spectra of data obtained from wear experiments using a 6-inch (10-blade) milling cutter. According to this, it can be seen that each frequency component increases little by little depending on the number of tests.

FIG. 17 shows data on the fluctuation width value and the average current when the feed amount per tooth and the feed speed are changed using the normal 5-inch milling cutter described above. Two cutting widths are used: 83mm and 17mm.
(Effect of feed rate) Figure 18 shows data when the depth of cut is 2 mm under the conditions of Figure 17 and the spindle rotational speed is varied. (Influence of Cutting Speed) FIG. 19 shows data when the depth of cut is changed under the conditions shown in FIG. 17. (Influence of depth of cut) What can be said from Figures 17 to 19 above is that even if the cutting speed and feed rate change, the average value and fluctuation range of the current do not change much, and the depth of cut and cutting width The point is that when the value changes, both the average value and the fluctuation range change.

Figure 20 shows No.1 to No.6 (No.1...normal blade, No.2
...The average current value and fluctuation range value readings obtained through a 10Hz low-pass filter were plotted when the depth of cut was changed for one chipped edge (see Figure 5 for No. 3 to No. 6). It is something.

Referring to the figure, it can be seen that only the variation width value corresponding to the chipped edge (No. 2) has an abnormally large change.

FIG. 21 shows data for investigating the influence of wear, and it can be seen that the fluctuation range value increases with the number of cuttings. It should be noted that in this case, the average current value also tends to increase.

For example, in 10 cuts, the variation range (reading)/average current (reading) is 2.8/28≒1/10, but in 60 cuts, the result is approximately 4.2/32>1/10. .

According to the experimental data presented so far, it has been found that when a cutting abnormality such as breakage or wear occurs, the fluctuation range value of the spindle motor current mainly changes. However, it is also true that the above-mentioned fluctuation range value changes greatly depending on changes in the depth of cut and cutting width, and it is not possible to detect breakage or the like simply from a change in the fluctuation range. However, in the data mentioned above, when the depth of cut or cutting width changes, not only the range of variation but also the average value changes, so the inventors paid attention to the mode of change and divided the range of variation by the average value. When we look at the dimensional quantity K, we find that it has a certain tendency or significance.

Figure 22 plots the amount K when the depth of cut is changed based on the data mentioned above for a 6-inch (10-blade) milling cutter. It can be seen that in the case of a chipped edge, the value of the amount K is considerably larger than that of the other plot points (No. 1, No. 3 to No. 6). Moreover, in the case of No. 1 normal blade and defective blades No. 5 and No. 6, K is relative to the depth of cut.
It is also understood that there is almost no change.
Also, the value K of chipped blade No. 3 remains unchanged at around 0.2.
Furthermore, for chipped blade No. 4, on the contrary, it decreased from close to 0.2 to below 0.1.

FIG. 23 is a graph showing the relationship between the feed amount and the amount K for a 5-inch (8-blade) milling cutter, and the plot marked with ◯ is for a normal blade. On the other hand, the plot points of one chipped edge marked with an x are also distributed upward.

FIG. 24 is a graph plotting the relationship between the spindle rotational speed (rpm) and K. The plot marked with ◯ indicates the case of a normal blade, and the plot marked with × indicates a case where one blade is missing.

In Figures 23 and 24, data was not taken for the plot points marked with an x in correspondence to a normal blade, but this is because in actual cutting, a cutting edge of less than 0.25mm/tooth is not used in Figure 23, and in Figure 24. This is because cutting is not performed when the spindle rotation speed is less than 220 rpm.

FIG. 25 plots the relationship between the number of cuts and K using a 6-inch (10-blade) milling cutter for a normal cutting edge, and shows the change in K due to the progress of wear.

It can be seen that the quantity K is gradually increasing according to the figure. What can be said from Figures 23 to 25 above is that by monitoring this dimensionless quantity K, it is possible to not only detect breakage (chipped blades) but also determine appropriate judgment values for wear. Detection is possible by setting.

In this case, one technical issue is which method to use for determination, but the method that the present inventors have implemented is first
The amount K when cutting in the state of Ti is set as Ko (Ti), and the value K (Ti) during cutting with that tool Ti is hereinafter used.
The ratio of Ko(Ti) is taken, and an abnormality in the tool Ti is determined based on whether the value of the ratio exceeds a certain set value α.

FIG. 26 is a block diagram of an embodiment of the present invention proposed based on the above-mentioned idea.

In the figure, a shunt circuit 62 detects the armature current of a motor, for example a DC motor 61, for driving the rotation of the main shaft on which a milling cutter is attached. 63 is a low pass filter whose output is converted into a digital value by an A/D converter.

Reference numeral 65 denotes a calculation unit for calculating the variation width value W of the cutting current value of the motor 61, which calculates the peak-to-peak value of the digital value for a certain period of time.

Reference numeral 66 denotes a motor current value average value (R) calculation unit, which calculates the average value of the current value at fixed time intervals during cutting. Reference numeral 67 denotes a calculation unit that calculates a value K of the ratio between the width value W and the average value R, and thereby calculates one dimensionless quantity K from the two quantities R and W.

Reference numeral 68 denotes a detection unit for detecting a stable state of the average value R, and provides a command signal CD so that the determination unit 71 performs a determination operation using the output K from the calculation unit 67. The detection unit 68 may have various configurations;
For example, each time the value (Ru) of the average value calculation unit 66 is calculated during cutting, the previous calculated value (Rn
-1), a stable state is defined as a state in which the difference is within a certain value (δ). That is, a stable state can be defined when the following expression is satisfied: |Rn−Rn−1|≦δ.

Reference numeral 69 is a gate for storing the output Ko of the calculating section 67 during cutting in the virgin state of the milling cutter currently in use in the storage section 70, and the above-mentioned command signal CD is applied to the gate 69. . The storage unit 70 stores tool numbers T and N.
When is given as an address command, a memory area corresponding to that address is secured and the value Ko is stored there. Here, the tool numbers T and N are not only distinguished by the shape of the tool, that is, the milling cutter, but also by the same cutter if the material of the workpiece is different.

However, no distinction is made with respect to changes in cutting conditions such as depth of cut and cutting speed.

In the presence of the command signal CD, the determination unit 71 calculates the ratio K/Ko between the output K of the calculation unit 67 and the value Ko of the corresponding milling cutter, which is called up from the storage unit 70 according to the specification of T and N at that time. Setting value αi according to the type of judgment set in advance
A determination signal i is generated depending on whether K/Ko>αi is satisfied. Here, αi = α1, α2, α3... and for example, if i = 1, the tool may be broken (chipped edge), or i
In the case of i=2, an abnormal state such as tool wear is determined, and in the case of i=3, a tool is damaged.

In FIG. 26, a DC motor is used as an example of the motor, but an AC motor may also be used. Also A/D converter 6
4, the width value W and the average value R are calculated, but the quantity K is calculated using the analog signal as it is, and an A/D converter is provided before the memory 70 and the determination section 71. Good too. Furthermore, an analog memory may be used as the memory, and the determination section 71 may also perform determination using an analog quantity adder K+(-Ko·αi>0).

Furthermore, in the above explanation, the spindle motor (DC, AC
However, the gist of the present invention can be realized even with other physical quantities or electrical quantities that reflect the cutting force. For example, it is also possible to use the motor current value of the drive system that moves the workpiece during cutting, or the servo lag of the drive control system.

As explained above, the present invention focuses on the value K of the ratio between the variation width value W of the cutting force during milling cutting and the average value R, and compares this with the value Ko of the ratio during virgin cutting of the milling cutter. Accordingly, it is possible to detect an abnormal state of the milling cutter regardless of changes in cutting conditions such as depth of cut, spindle rotation speed, etc. during milling. In particular, by calculating W and R using a digital method as shown in FIG. 26, it is also possible to perform software processing on the calculation section 67, determination section 71, detection section 68, etc., including the memory 70.
It can also be installed as a simple optional device on existing machine tools.

[Brief explanation of the drawing]

Figure 1 is a block diagram of a measurement system for measuring various data associated with cutting, and Figure 2 is a diagram showing the locus of the center of the milling cutter (6-inch, 10-blade) as it moves relative to the workpiece. , Figure 3 is a diagram showing the locus of the cutter center in the case of a 5-inch (8-blade) cutter, Figure 4 is a diagram showing how the workpiece is attached to the table, and Figure 5 is a diagram showing the trajectory of the cutter center in the case of a 5-inch (8-blade) cutter. The four corners of the blade were damaged with an electric pen to form a chipped edge at each corner. Figure 6 shows cutting with a 6-inch (10-blade) milling cutter obtained through a 10Hz low-pass filter. In the current waveform diagrams in the middle, a shows the current waveform for a normal blade, b shows the case for one chipped blade, Figure 7 shows the current waveform for the chipped blade under the same cutting conditions as Figure 6, and c shows the current waveform for the chipped blade. Missing blade (No. 3), d is a diagram showing the missing blade (No. 4), No. 8
The figure shows the same cutting conditions as in Figure 6, e is the current waveform during cutting by the chipped blade (No. 5), f is the current waveform during cutting by the chipped blade (No. 6), and Figures 9a and b are the current waveforms for the chipped blade (No. 6). Figures corresponding to Figures a and b show the current waveform obtained through a 50Hz low-pass filter, Figure 10 shows the power spectrum of the current when the 6-inch milling cutter has a normal blade, and Figure 11 shows the current waveform obtained through a 50Hz low-pass filter. Figure 12 shows the power spectrum of current when a 6-inch milling cutter has one chipped blade, Figure 12 shows the power spectrum of current when a 5-inch milling cutter has one chipped blade, and Figure 13 shows the power spectrum of the current when a 5-inch milling cutter has one chipped blade. Figure 1 shows the power spectrum of the current when the milling cutter has one missing blade and the cutting width is different from that in Figure 12.
Figure 4 shows the power spectrum of the current when changing the spindle rotation speed, Figures 15 and 16 show the power spectrum of the current for detecting the effects of wear, and Figure 17 shows the feed amount. Figure 18 is a diagram showing the current fluctuation range and average value change due to spindle rotation (cutting speed), Figure 19 is a diagram showing the current fluctuation range and average value change due to spindle rotation (cutting speed), and Figure 19 is a diagram showing the current fluctuation range and average value change due to spindle rotation (cutting speed). Figure 20 shows the variation width and average value of current depending on the depth of cut. Diagram showing changes,
Figure 21 shows the effects of wear,
A diagram showing the current fluctuation range and average value depending on the number of cuttings,
Fig. 22 shows the change in value K (fluctuation range/average value) depending on the depth of cut, Fig. 23 shows the change in value K depending on the feed amount, and Fig. 24 shows the change in value K depending on the spindle rotation speed. FIG. 25 is a diagram showing changes in the value K depending on the number of cuttings, and FIG. 26 is a block diagram of a cutting abnormality detection device according to the present invention. 61... Motor, 62... Shunt circuit, 63... Low pass filter, 64... A/D converter, 65... Width value calculation section, 66... Average value calculation section, 67... Calculation section, 6
8... Average value stable state detection unit, 69... Gate, 70
... Storage section, 71... Judgment section.

Claims (1)

[Scope of Claims] 1. A device for detecting cutting abnormality of a cutting tool during milling, comprising: a cutting force detection means for detecting a physical quantity corresponding to the cutting force during the cutting; and an intermittence of the cutting tool during the milling. Width value calculation means for calculating a fluctuation width value (W) of the output of the cutting force detection means due to cutting; Average value calculation means for calculating an average value (R) of the output of the cutting force detection means; stable state detection means for detecting that the cutting tool is in a stable state; calculation means for calculating the ratio between the width value W and the average value R; and the width value measured during virgin cutting with the cutting tool. When there is an output from the stable state detection means, the storage means stores the value of the ratio of Wo and the average value Ro, and when there is an output from the stable state detection means, the W/R and Wo/Ro are compared, and the value of the ratio is determined in advance. A cutting abnormality detection device for milling cutting, characterized in that it comprises a comparison and determination means for determining that the cutting tool is abnormal when a predetermined constant value is exceeded. 2. A cutting abnormality detection device in milling cutting, characterized in that the cutting force detection means in the device according to claim 1 uses a value of current flowing through the spindle rotation drive motor as a detection symbol. 3. A cutting abnormality detection device in milling according to claim 2, characterized in that the spindle drive motor current value is detected through a low-pass filter of about 10 Hz. 4. A cutting abnormality detection device in milling according to claim 3, characterized in that the current value is detected via an A/D converter.