METHOD AND DEVICE FOR DETERMINING ROTATIONAL SLIPPAGE OF A TOOL FRICTIONALLY HELD IN A TOOL HOLDER

Abstract

A method and a device for determining a rotational slippage of a tool that is frictionally held in a tool holder. The rotational slippage of the tool is determined by evaluating a phase and/or a period length of a periodic process signal from the tool.

Claims

1. A method for determining rotational slippage of a tool that is frictionally held in a tool holder, the method comprising: measuring a phase and/or a period length of a periodic process signal from the tool that is frictionally held in the tool holder; and determining a rotational slippage of the tool by evaluating at least one of a phase or a period length of the periodic process signal.

2. The method for determining the rotational slippage according to claim 1, wherein the periodic process signal is an electrical signal.

3. The method for determining the rotational slippage according to claim 1, wherein the periodic process signal is at least one signal selected from the group consisting of a spindle current, a force signal, an acceleration signal, an optical signal, an acoustic signal, and a vibration signal.

4. The method for determining rotational slippage according to claim 1, wherein the tool is a machining tool.

5. The method for determining rotational slippage according to claim 4, wherein the machining tool is a tool selected from the group consisting of a milling tool, a turning tool, a grinding tool, and a boring tool.

6. The method for determining rotational slippage according to claim 5, wherein the machining tool is an end mill.

7. The method for determining rotational slippage according to claim 1, wherein the tool holder is a chuck.

8. The method for determining rotational slippage according to claim 7, wherein the chuck is selected from the group consisting of a shrink chuck, a collet chuck, and a hydro expansion chuck.

9. The method for determining rotational slippage according to claim 1, which comprises further processing the periodic process signal by filtering, smoothing, standardizing, and/or eliminating a trend of the periodic process signal.

10. The method for determining rotational slippage according to claim 1, which comprises comparing the periodic process signal with an artificially generated periodic reference signal, and including a periodically repeating event in the periodic process signal as a periodically repeating event in the artificially generated periodic reference signal, and comparing at least one of the phase, the period length in the periodic process signal, or the periodic reference signal.

11. The method for determining rotational slippage according to claim 10, which comprises, when the comparing step indicates a change over time, inferring rotational slippage.

12. The method for determining rotational slippage according to claim 10, which comprises carrying out the step of comparing the periodic process signal with the periodic reference signal in a frequency domain.

13. The method for determining rotational slippage according to claim 12, which comprises carrying out the comparing step by mapping the periodic process signal and the periodic reference signal in the frequency domain by using a Fourier transform.

14. The method for determining rotational slippage according to claim 10, which comprises, during a comparison of the periodic process signal with the periodic reference signal, determining a phase difference between the periodic process signal and the periodic reference signal and inferring rotational slippage from a change in the phase difference.

15. The method for determining rotational slippage according to claim 10, which comprises initializing an offset of the periodic reference signal such that, at a start of the comparison, a phase difference between the periodic process signal and the periodic reference signal is equal to zero.

16. The method for determining rotational slippage according to claim 9, wherein the periodic process signal is a spindle current or a force signal or a vibration signal, and the periodically repeating event on which the periodic process signal is based is a tooth engagement.

17. The method for determining rotational slippage according to claim 1, which comprises determining a tool pull-out of a tool mounted in a tool holder and determining the tool pull-out from a rotational slippage by using a pull-out twist angle.

18. The method for determining rotational slippage according to claim 1, wherein, when the determined rotational slippage exceeds a predefined threshold value, aborting a process and/or changing its process parameters, or compensating a tool pull-out determined from the rotational slippage.

19. A device for determining rotational slippage of a tool that is frictionally held in a tool holder, the device comprising an evaluation unit configured to carry out the method according to claim 1.

20. The device for determining rotational slippage according to claim 19, further comprising sensors configured to detect the periodic process signal disposed on at least one of a machine tool, a tool holder, or a tool.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0059] FIG. 1A shows a perspective view of a machine tool with a tool that is frictionally held in a tool holder according to an embodiment of the invention;

[0060] FIG. 1B shows the tool holder separately;

[0061] FIG. 1C shows the tool separately;

[0062] FIG. 2 shows illustrations which illustrate a geometric relationship between a tool pull-out and a phase shift; and

[0063] FIG. 3 shows phase variations of periodic process signalswith and without tool pull-out.

DETAILED DESCRIPTION OF THE INVENTION

[0064] First, we shall describe the data-driven identification of slippage/tool pull-out of a (rotating) frictional tool-tool holder connection during machining:

[0065] FIG. 1A shows a machine tool 2, here a milling machine 2, having a tool 6 frictionally held in a tool holder 4, here in a collet chuck 4 (illustrated in more detail and separately in FIG. 1B), here an end mill 6 (illustrated in more detail and separately in FIG. 1C) (with tool cutters N=5 and tool diameter D=16 mm), in order to carry out machining on a workpiece (not illustrated), for example made of a titanium alloy.

[0066] The machine tool 2 is also equippedwith an appropriate, i.e., appropriately set up, evaluation unit 8in order to detect and to monitor slippage or tool pull-out during the machining.

[0067] The monitoring or identification of the slippage, i.e., the rotational and axial slippage, or tool pull-out is carried out in a data-driven mannerby using a (periodic) process signal (detected by measurements during the machining), i.e., in this case of the spindle current y.sub.Sp of the machine tool 2 that can be tapped off in the internal machine controller 10 (or is present as data there).

[0068] Now we shall describe the mathematical principles of the monitoring:

Example 1: Analysis Via a Comparison Between Process Signal and (Artificially Generated) Reference Signal

[0069] Within the context of pre-processing (data pre-processing), firstly the trend of the periodic profile in the process signal or in the spindle current y.sub.Sp is eliminated. The process signal is then standardized, so that the amplitude profile present is as far as possible process parameter-independent.

[0070] A reference signal y.sub.ref(t)to be compared with the process signalcan be generated (artificially) from the known process parameterscutting speed v.sub.C, tool diameter D and number of tool cutters N.

[0071] After adding the auxiliary variables rotational speed n (n=?.sub.C/?*D) and tooth engagement frequency f.sub.Tooth (f.sub.Tooth=N*n), the result is a simple sinusoidal profile for the reference signal y.sub.ref(t):

[00001]$y_{\mathrm{ref}}\left(t\right)=\sin \left(2?.Math.f_{\mathrm{Tooth}}.Math.t+{?}_{\mathrm{ref},0}\right)$

[0072] Here, the offset ?.sub.ref,0 is firstly initialized, so that the phase difference ?.sub.diff between the reference signal ?.sub.ref and the process signal or the spindle current ?.sub.Sp is equal to zero at the start of the monitoring time.

[0073] Once the generation of the artificial reference signal has been carried out, then it is finally possible to infer the phase difference ?.sub.diff(t): Now, the signals over the entire time period (of a process to be monitored) are mapped in the time-frequency domain by means of a short-time Fourier transformation and then compared by using the tooth engagement frequency:

[00002]${?}_{\mathrm{diff}}\left(t\right)={?}_{\mathrm{ref}}\left(t\right)-{?}_{\mathrm{Sp}}\left(t\right)$

[0074] If slippage or rotational and axial slippage and an increasing tool pull-out leads to a higher phase shift and phase difference, then the slippage or the tool pull-out during the machining can be identified and monitored by using the phase difference (as illustrated below by way of example for (trial) processes).

[0075] Adjustable maximum permissible phase differences (corresponding to maximum permissible slippages or maximum permissible tool pull-outs) can be stored in the internal machine controller 10 and (during real-time monitoring) lead to the machine tool 2 being automatically switched off.

Example 2: Analysis of the Process Signal

[0076] An alternative approach is based on time series data with a signal size of N, which is a k multiple of the data points acquired during one spindle revolution.

[0077] According to equation (1), this can be calculated by multiplying the number of k rotations by the ratio of sampling frequency fs and rotational frequency f.sub.rot.

[00003]$\begin{array}{cc}N=k.Math.\frac{f_s}{f_{\mathrm{rot}}}& \mathrm{Eq}.\left(1\right)\end{array}$

[0078] As a precondition, this signal must be sensitive enough to detect the periodic tooth engagement. The sensitivity can be quantified by using the signal to noise ratio of the signal amplitude at the tooth engagement frequency, which must be at least greater than 0 dB.

[0079] The tooth engagement frequency f.sub.t can be described as an n.sub.t multiple of the rotational frequency f.sub.rot, where n.sub.t is the number of cutting edges.

[00004]$\begin{array}{cc}{f}_t=n_t.Math.f_{\mathrm{rot}}& \mathrm{Eq}.\left(2\right)\end{array}$

[0080] Firstly, the time series signal is transformed into the spectral domain in order to ensure improved representation of the signal amplitude at the tooth engagement frequency. For a given discrete time domain signal y.sub.n, the discrete Fourier transformation is defined as follows:

[00005]$\begin{array}{cc}Y\left(f_k\right)={.Math.}_{n=0}^{N-1}{y}_n.Math.e^{-\frac{i2?}{N}{f}_k.Math.n}& \mathrm{Eq}.\left(3\right)\end{array}$

[0081] The transformed frequency domain signal Y(f.sub.k) then consists of complex data points, which can be broken down into their magnitude and their phase:

[00006]$\begin{array}{cc}Y\left(f_k\right)=\mathrm{Re}\left\{Y\left(f_k\right)\right\}+i.Math.\mathrm{Im}\left\{Y\left(f_k\right)\right\}& \mathrm{Eq}.\left(4\right)\end{array}$$\mathrm{where}$$\begin{array}{cc}\hat{y}\left(f_k\right)=.Math.Y\left(f_k\right).Math.=\sqrt{\mathrm{Re}{\left\{Y\left(f_k\right)\right\}}^2+\mathrm{Im}{\left\{Y\left(f_k\right)\right\}}^2}& \mathrm{Eq}.\left(5\right)\end{array}$$\mathrm{and}$$\begin{array}{cc}?\left(f_k\right)=\arg \left(Y\left(f_k\right)\right)={\tan }^{-1}\frac{\mathrm{Re}\left\{Y\left(f_k\right)\right\}}{\mathrm{Im}\left\{Y\left(f_k\right)\right\}}& \mathrm{Eq}.\left(6\right)\end{array}$

[0082] Eq. (5) and Eq. (6) with ? and ? are designated as the amplitude spectrum and phase spectrum. The signal size at the tooth engagement frequency ?(f=f.sub.t) is viewed as a measure of the signal sensitivity in relation to the tooth engagement, while ?(f=f.sub.t) represents the primary measure for the detection of the tool rotational slippage or pull-out.

[0083] Therefore, the phase difference between the current phase of the last obtained data packet oi and the phase of the first obtained data packet do is calculated continuously in order to quantify the rotation during the process.

[00007]$\begin{array}{cc}\mathrm{??}=.Math.{?}_i-{?}_0.Math.& \mathrm{Eq}.\left(7\right)\end{array}$

[0084] The ratio ??/n.sub.t between the phase difference ?? and the number of cutters n.sub.t describes the angular displacement of the relative rotation between tool and tool holder (circumferential angle, rotational slippage).

[0085] However, in order to obtain the new tool offset, which is correlated with the relative axial movement, further calculations have to be carried out.

[0086] Therefore, the circumferential angle of the rotational movement is firstly converted into the arc length Au, as illustrated in FIG. 2, which in turn depends on the tool diameter Dt:

[00008]$\begin{array}{cc}?u=\frac{\mathrm{??}}{n_t}.Math.\frac{?}{360?}.Math.D_t& \mathrm{Eq}.\left(8\right)\end{array}$

[0087] Nextas illustrated in FIG. 2? is introduced as the (tool) pull-out twist angle ? of the helical movement during tool pull-out. The relative axial tool displacement can be determined therefrom as follows:

[00009]$\begin{array}{cc}?l=\frac{?u}{\tan ?}& \mathrm{Eq}.\left(9\right)\end{array}$

[0088] Now we shall describe the verification of the monitoring by using trial processes (cf. Table 1 and FIG. 3)

[0089] The process signalfor the processes to be monitoredoriginates from milling trials on the machine tool 2 when processing the workpiece 6 by synchronous milling.

[0090] The tool 6 was fixed in the tool holder 4 via the union nut 12 by means of a torque wrench (in trials (a) with a higher tightening torque (here 136 Nm) and (b) with a lower tightening torque (here, 56-80 Nm)), as a result of which the tool 6 was mounted frictionally in the tool holder 4once more frictionally (in (a)) and once less frictionally (in (b)).

[0091] The different tightening torques of the union nut 12 (according to (a) and (b)) in turn lead to a differently formed press joint between the tool 6 and tool holder 4. A reduced tightening torque, as according to the trials (b), leads to a weakly pronounced press joint pressure and thus to a component connection that is prone to slippage and pull-out.

[0092] The cutting parameters illustratedtooth advance f.sub.z (cf. table 1, in the columns there), and radial width of cut de (cf. table 1, in rows there)are chosen by using prior trials such that pull-out can be prevented under control or permitted.

[0093] After each trial, the tool length of the tool 6 is measured by means of a laser sensor (not illustrated). By means of referencing to the initial length (before the trial), the tool pull-out present can ultimately be determined. This is in each case entered in the cells of table 1.

[0094] Table 1 thus shows an overview of the pull-out trials (a) and (b) carried out on the machine tool 2. The cutting speed ?.sub.C and the axial width of cut a.sub.p in each case remain unchanged at 80 m/min and 45 mm.

TABLE-US-00001 TABLE 1 Tool pull-out in axial Tooth advance f.sub.z in mm direction 0.06 0.08 Radial width of 1 0 ?m 0 ?m cut a.sub.e in mm 1.5 0 ?m 0 ?m Radial width of 1 775 ?m 924 ?m cut a.sub.e in mm 1.5 1025 ?m (a) Tightening torque of union nut: 136 Nm (maximum) (b) Tightening torque of union nut: 56-80 Nm (reduced)

[0095] FIG. 3 shows by way of example the phase difference profiles highlighted (in bold) in table 1, determined from spindle current data of the internal machine control data of the internal machine controller 8 during the respective linear milling operation.

[0096] It can be seen from FIG. 3 that the profiles of the phase difference without any tool pull-out (pull-out=0 ?m) remain constant, while the phase difference profiles with tool pull-out (pull-out=775 ?m and pull-out=924 ?m) exhibit a falling trend.

[0097] This means it was possible to verify the fact that an increased tool pull-out leads to a higher phase shift. This causal relationship was confirmed statistically by repeated trials.

[0098] The data-driven approach saves costly constructional measures during the detection or avoidance or compensation of the tool pull-out and offers a more universal solution, since process parameter-independent monitoring or control is possible.

[0099] The requirements on the implementation are low; it is merely necessary to detect a periodic process signal, such as the spindle current here, by using which the tooth engagement frequency can be detected.

[0100] Trials (with the spindle current) have shown that even internal machine data, if available, is sufficient for the purpose, and no additional measuring device is required.

[0101] In addition, the analytical method is stable against interference and noise, since these signal components are filtered out by a bandpass filter. Because of the real-time capability, complete pull-out compensation is potentially conceivable.

[0102] Although the invention has been illustrated and described in more detail through the preferred exemplary embodiments, the invention is not restricted by the examples disclosed and other variants can be derived therefrom without departing from the protected scope of the invention.

[0103] The following is a summary list of reference numerals and the corresponding structure used in the above description of the invention: [0104] 2 Machine tool [0105] 4 Tool holder, chuck, collet chuck [0106] 6 Tool, milling tool, end mill [0107] 8 Evaluation unit [0108] 10 Internal machine controller [0109] 12 Union nut