METHOD FOR GRINDING A TOOTH OR PROFILE OF A WORKPIECE

Description

[0001] The invention relates to a method for grinding a gear or a profile of a workpiece by means of a grinding tool in a grinding machine, wherein the grinding tool is mounted on a tool spindle and the tool spindle is rotated by means of a first drive motor, wherein the workpiece is mounted on a workpiece spindle and the workpiece spindle is rotated by means of a second drive motor, wherein at least during the engagement of the grinding tool in the gear to be ground or in the profile to be ground, a first and a second machine parameter are measured and both measured machine parameters or a variable derived from the machine parameter is compared with a predetermined stored value, wherein a signal is output if at least one of the machine parameters or the variable derived therefrom exceeds or falls below the predetermined stored value, taking into account a tolerance band.

[0002] Such a method according to the preamble of independent claim 1 is known from WO 2020 / 193228 A1. Here, monitoring of the type described is carried out to detect whether a grinding wheel breakage has occurred in a grinding worm. If so, a warning indicator is issued indicating an impermissible process deviation.

[0003] When monitoring the grinding process, it is generally known to observe a relevant process parameter, such as the current consumption of the motor driving the grinding spindle, and check whether the value lies within a specified tolerance band over the course of a grinding stroke. This tolerance band naturally changes over the course of the grinding stroke. If the value leaves the tolerance band, a signal is issued, indicating an improper process condition. The specified tolerance band can therefore be understood as a range of values ​​that has been "taught" by the system and draws on past experience. For the state of the art in this regard, reference is made to the tool monitoring system from Nordmann GmbH & Co. KG, Hürth, Germany (see: https: / / www.nordmann.eu / huellkurven.html).

[0004] While this makes it possible to gain insight into the progress of the grinding process, particularly with regard to whether a part was ground correctly, it has been found that in many cases this is not sufficient to make a sufficient statement about the progress of the grinding process and the quality of the ground part. In particular, it is difficult to draw conclusions about the cause of the defect if the grinding process was not carried out correctly.

[0005] The invention is based on the TaskThe aim is to further develop a generic method in such a way that it is possible to obtain improved information about the progress of the grinding process by monitoring relevant variables. This should enable a better assessment of whether a part has been ground correctly. If necessary, it should also be possible to derive indications of the problems that caused the grinding process to fail. Thus, one aim of the present invention is to ensure reliable monitoring of the grinding process and the most complete monitoring of the workpiece quality possible, which should be possible quickly and cost-effectively and thus without subsequent external measurement of the ground part, in particular the gear.

[0006] The solution to this problem by the invention provides that at least one of the machine parameters or the variable derived from this machine parameter contains periodic signal components, wherein these signal components are broken down into the individual frequency components by a frequency analysis and the frequency components are used for comparison with regard to their frequency and / or their amplitude.

[0007] Preferably, both machine parameters or the quantities derived from these machine parameters contain periodic signal components.

[0008] Preferably, the frequency components are used for comparison only with regard to their amplitude.

[0009] According to a preferred approach, the frequency analysis is carried out by means of a fast Fourier transform (FFT).

[0010] However, alternative and well-known methods can also be used for this purpose, in particular a discrete Fourier transform (DFT), a root-mean-square analysis (determination of the RMS spectrum), a determination of the amplitude spectrum, a cepstrum analysis, a compensating sine function, or a determination of the auto-power spectrum (PSD analysis). These signal analysis methods are well-known per se and therefore need not be discussed in detail here.

[0011] The measurement of the first and second machine parameters can be carried out during a specified time interval while the workpiece is being ground with the grinding tool.

[0012] It is also possible for the first and second machine parameters to be measured over a specified feed path while grinding the workpiece with the grinding tool. These parameters (in particular, the grinding performance of the tool spindle) are therefore not observed temporally, but spatially (over the course of the grinding stroke). This makes it possible to examine particularly relevant sections and compare them with previously stored data.

[0013] According to a preferred embodiment of the invention, the first machine parameter is the current consumption of the motor of the tool spindle. The second machine parameter is preferably the current consumption of the motor of the workpiece spindle.

[0014] Regarding the power or current consumption of the tool motor, a further development of the invention provides that the current consumption curve resulting from a grinding stroke without material removal (or from a finishing stroke, which consumes comparatively little energy) is subtracted from the total current consumption of the motor. Therefore, this approach does not consider the entire spindle power, but only the cutting power. The portion resulting from an idle stroke (or a finishing stroke) is deducted from the total power consumption.

[0015] During the idle stroke (or the finishing stroke), only the power loss is recorded, such as that caused by the bearings of the axes or by hydrodynamic losses due to the cooling lubricant.

[0016] When we talk about the current consumption (I) of the motors, we also mean the power consumption (P) of the motors, which is determined by the relationship P = p ‾ = 1 T ∫ t 0 t 0 + T u ⋅ i d t when the voltage U is applied.

[0017] One of the recorded machine parameters can also be the structure-borne sound of the grinding machine or a part of it, whereby the structure-borne sound is recorded via a structure-borne sound sensor.

[0018] An alternative is that the quantity derived from the machine parameter is the cutting energy per volume (measured in J / mm 3< ) required to machine a given volume of the stock to be removed from the gearing or profile. The power or current consumption of the tool spindle can be used to determine the energy absorbed for a defined period of time that is required to remove stock (particularly after deducting the components that are not attributable to cutting, see above). The corresponding machined volume can then be determined taking into account the geometry of the tool (grinding worm) and the gearing or profile as well as the process parameters (particularly the infeed of the tool onto the workpiece).The mentioned parameter of cutting energy per volume facilitates the transferability between different grinding processes (especially generating grinding processes) for different workpieces.

[0019] From the measured machine parameters and / or from the variable derived from the machine parameter, a characteristic value can be determined and output that characterizes the grinding process for the ground workpiece.

[0020] The grinding tool is preferably a grinding worm and the workpiece a gear.

[0021] The proposed concept therefore envisages monitoring and evaluation of the grinding process, particularly the generating grinding process, by combining multiple signals, taking into account at least two different process parameters. For each ground workpiece, a characteristic value can be defined that provides information about the grinding process being performed.

[0022] The described procedure effectively monitors the grinding process and detects defective or abnormal components at an early stage. In particular, defects in the workpiece blank, waviness on the flanks of the ground gear, and tool defects are detected.

[0023] The assessment is performed by accessing data stored (in the machine control system) and thus learned insights from previous grinding processes. This allows process deviations (anomalies) to be detected more effectively, allowing the machine operator to be warned or the grinding process to be aborted.

[0024] During machining, several internal control signals (i.e. those present in the machine control system) and also external control signals (e.g. recorded via sensors that record structure-borne noise, which originates, for example, from the machine bed or the hall floor) are recorded and taken into account.

[0025] In addition, machine-internal data (such as set corrections, the diameter of the grinding worm, generated paths along which the workpiece and the tool are guided relative to each other) can be used to evaluate the process, for which purpose they can be adaptively filtered and sorted if necessary (for example, by dividing the entire grinding process into different strokes, dividing it into infeed, outfeed and full engagement of workpiece and tool).

[0026] Depending on the factors influencing the process, especially with regard to the generated path (influenced by screw diameter and corrections), characteristic values ​​can be calculated and output.

[0027] In addition to the previously known methods (see above), characteristic values ​​for the aforementioned subdivision (statistical parameters such as standard deviation, variance, kurtosis, skewness, frequency distributions, percentile classification, slope, area) can be determined from the individual recorded signals; this also applies to the determination of characteristic values ​​from signal analysis (such as frequency, amplitude, order analysis).

[0028] To ensure the general validity of the determined data for different grinding processes and thus for transferability, the above-mentioned parameters can be standardized. For this purpose, for example, the relative metal removal rate, the relative grinding performance, the relative cutting force, and the contact time between the workpiece and the tool are used.

[0029] Information about the grinding process is generated by combining the various parameters obtained from the various sensor signals.

[0030] Using algorithms from statistics, and especially from the field of machine learning, this data is evaluated, thereby determining the quality of the processing. The well-known algorithms from the field of machine learning include both supervised and unsupervised learning, "deep learning," and "reinforcement learning."

[0031] This makes it possible to improve monitoring quality and thus stabilize the grinding process. It is easier to detect whether a process is faulty, and the type of error can also be identified more accurately. The prerequisite for this is that such an error or a similar error has been learned into the data stored (in the machine control system).

[0032] In addition to detecting the error, this also enables faster troubleshooting. Based on this knowledge, the machine is able to intervene adaptively in the process and optimize it.

[0033] As mentioned, based on insights into the grinding process, the measurement signals in the process can be divided into areas that are meaningful for evaluating the process. Thus, not the entire machining process is necessarily assessed, but only relevant areas.

[0034] The proposed procedure thus monitors the grinding process and evaluates it based on an index. This means that for each individual workpiece, an index, i.e. a numerical value, is calculated which assesses the quality of the grinding process. These index values ​​can then be made available to the machine operator as a progress diagram, allowing them to easily obtain an overview of the grinding process. Several key values ​​can be used to determine the aforementioned index. These key values ​​then evaluate the grinding process, the dressing process, and the alignment of the workpiece in the grinding machine in different ways. This allows errors to be detected early on that would otherwise only be detected in the measuring room or not at all. These include errors such as raw part defects or setup errors.

[0035] Additionally, it is possible to issue targeted messages about irregularities if they have been learned into the algorithm. This allows for process optimization and error correction in a targeted and time-saving manner.

[0036] The data recorded during the specific grinding process is compared with data stored in a database. This allows, for example, specific process signals to be assigned to specific areas of the grinding worm; correspondingly, signals can be assigned to specific worm diameters and corresponding grinding worm speeds.

[0037] The drawing shows embodiments of the invention. Fig. 1 initially shows in a general way for a first grinding process in the upper figure the curve of the power consumption or the current consumption of the motor that drives the grinding spindle, and in the lower figure the curve of the power consumption or the current consumption of the motor that drives the workpiece spindle, and Fig. 2 shows according to an embodiment of the method according to the invention for a second grinding process in the left figure the curve of the power consumption or the current consumption of the motor that drives the workpiece spindle, and in the right figure the amplitudes of the frequency components of a periodic curve of the power or the current obtained from a Fast Fourier Transformation (FFT).

[0038] In Figure 1The curve of the power P or current I is plotted against time t, which results for a motor driving a grinding spindle (S) and for the motor driving a workpiece spindle (W) in a gear grinding machine (shown for a roughing stroke in which the workpiece is ground to a high quality). Figure 1 The curve of the power PS is plotted over time as it is absorbed by the motor of the grinding spindle; the same curve is obtained for the current IS of the motor via the applied motor voltage.

[0039] The same applies to the lower illustration in Figure 1 , where the same information is given for the motor driving the workpiece spindle.

[0040] The power and current curves expected during the production of proper gears are shown by the corresponding dashed lines. Dashed lines above and below the dashed lines indicate the permissible tolerance bands for the power and current curves. EL marks the area where a grinding worm enters the gear to be ground, VS marks the full cut where the actual grinding takes place, and AL marks the area where the tool exits the workpiece.

[0041] The solid line indicates the actual power or current curve when grinding a specific workpiece.

[0042] It can be seen that the power and current consumption at the grinding spindle are following a normal curve, which would indicate proper grinding. However, this is not the case. Due to a flank form error (namely, an opposing tooth line deviation between the right and left flanks), the grinding spindle consumes the expected current, but not the workpiece spindle, whose power consumption lies outside the expected range, as can be seen from the lower illustration in Figure 1 results.

[0043] It should therefore be noted that, depending on the position or orientation of the flank form error, the power consumption or current consumption at the grinding spindle may well still be within the permissible tolerance band, while only the simultaneous consideration of the power consumption or current consumption at the workpiece spindle shows that there were problems with the pre-machining of the gear and that a corresponding flank form error is present.

[0044] In the present embodiment, it is generally intended that (at least) during the engagement of the grinding tool in the gear to be ground, a first machine parameter in the form of the power or current of the tool spindle and a second machine parameter in the form of the power or current of the workpiece spindle are measured. The two measured machine parameters PS / IS, PW / IW are compared with a predetermined stored value. Figure 1As shown in the figure below, it can be immediately seen that the power or current curve has left the permissible tolerance range, triggering a warning signal. The machine control system thus indicates a problem with the grinding process and that the part may not be good.

[0045] In Figure 2 a concrete procedure according to the invention is shown, according to which the method mentioned here has been specifically designed in such a way as to be able to make problems during the grinding process more visible and to issue a corresponding warning message.

[0046] In the left illustration in Figure 2 The power and current consumption at the workpiece spindle is again shown (see the solid line), which lies within the permissible tolerance range. However, it can also be seen that the power and current consumption changes periodically, i.e., it exhibits the characteristics of a superimposed oscillation.

[0047] This superimposed vibration can impose a detrimental fine waviness on the tooth flank, which can lead to noise during use of the gearing, which is particularly disadvantageous in the field of electromobility.

[0048] According to a possible embodiment of the invention, it is therefore provided here that in refinements of the above procedure, a variable derived from the machine parameter (here: from the power or current consumption of the workpiece spindle) is considered in order to assess or evaluate the grinding process.

[0049] The recorded course in the left part of Figure 2 is subjected to a Fast Fourier Transformation (FFT) to decompose the periodic signal into its components ("harmonics"). This is shown in the right part of the image in Figure 2Here, the amplitude A of the individual frequency components of the recorded periodic signal is schematically plotted over the order Or. It can be seen that one frequency component lies above a limit value Gr, so it can be concluded that a proper grinding process has not taken place.

[0050] It should be noted that the representation in Figure 2 This is only very schematic and, of course, individual limit values ​​can be set for each individual frequency component resulting from the FFT. Expert knowledge stored in the machine control system can be particularly useful here, causing it to issue a warning signal in the appropriate case.

[0051] The proposed approach makes it possible, in particular, to quickly identify improper process conditions before an entire batch of workpieces is potentially produced incorrectly. This can result in significant cost savings.

[0052] Of course, other parameters can also be used to monitor the process, with particular emphasis on recording the structure-borne noise of the machine or the hall floor, which, particularly after analysis of the recorded signals (for example using an FFT), can also provide information about how the grinding process progressed and whether it can be expected that a good part was ground.

[0053] The example illustrates the use of an FFT. Alternatively, any other known frequency analysis method can be used, such as, in particular, the Discrete Fourier Transform (DFT), frequency analysis using root-mean-square analysis (determining the RMS spectrum), determining the amplitude spectrum, cepstrum analysis (including variants such as power cepstrum), a compensating sine function, or determining the auto power spectrum (PSD analysis). All of the methods described are well-known in measured value analysis, so they need not be discussed in detail here.It is only essential that the individual frequency components of the measured periodic signal components are determined by means of a frequency analysis and that the results obtained from this are used for comparison with permissible limit values ​​(in particular for the maximum permissible values ​​of the individual amplitudes of the "harmonics").

Claims

1. Method for grinding a toothing or a profile of a workpiece by means of a grinding tool in a grinding machine, wherein the grinding tool is received on a tool spindle and the tool spindle is rotated by means of a first drive motor, wherein the workpiece is received on a workpiece spindle and the workpiece spindle is rotated by means of a second drive motor, wherein at least during the engagement of the grinding tool in the toothing to be ground or in the profile to be ground, a first and a second machine parameter (PS, PW) are measured and both measured machine parameters (PS, PW) or a variable (A) derived from the machine parameter (PS, PW) are compared with a predetermined stored value, wherein a signal is output if at least one of the machine parameters (PS, PW) or the variable (A) derived therefrom, taking into account a tolerance band, exceeds or falls below the predetermined stored value, characterized in that at least one of the machine parameters (PS, PW) or the variable (A) derived from this machine parameter (PS, PW) contains periodic signal components, wherein these signal components are broken down into the individual frequency components by a frequency analysis and the frequency components being used for comparison with regard to their frequency and / or their amplitude.

2. Method according to claim 1, characterised in that both machine parameters (PS, PW) or the variables (A) derived from these machine parameters (PS, PW) contain periodic signal components.

3. Method according to claim 1 or 2, characterised in that the frequency components are used for comparison only with regard to their amplitude.

4. Method according to one of claims 1 to 3, characterised in that the frequency analysis is performed by means of a Fast Fourier Transform (FFT).

5. Method according to one of claims 1 to 3, characterised in that the frequency analysis is performed by means of a Discrete Fourier Transform (DFT).

6. Method according to one of claims 1 to 3, characterised in that the frequency analysis is performed by a root mean square analysis (determination of the RMS spectrum) or by a determination of the amplitude spectrum or by a cepstrum analysis or by a compensation sinus function or by a determination of the auto power spectrum (PSD analysis).

7. Method according to one of claims 1 to 6, characterised in that the measurement of the first and second machine parameters (PS, PW) is performed during a predetermined time interval during grinding of the workpiece with the grinding tool.

8. Method according to one of claims 1 to 6, characterised in that the measurement of the first and second machine parameters (PS, PW) is carried out over a predetermined feed path during grinding of the workpiece with the grinding tool.

9. Method according to one of claims 1 to 8, characterised in that the first machine parameter (PS) is the power or current consumption of the motor of the tool spindle.

10. Method according to one of claims 1 to 9, characterised in that the second machine parameter (PW) is the power or current consumption of the motor of the workpiece spindle.

11. Method according to claim 9 or 10, characterised in that from the course of the total power or current consumption of the motor the course of the power or current consumption is subtracted which results in a grinding stroke without cutting of material or only with low cutting in a finishing grinding stroke.

12. Method according to one of claims 1 to 8, characterised in that one of the machine parameters is the structure-borne sound of the grinding machine or a part thereof, which is detected via a structure-borne sound sensor.

13. Method according to one of claims 1 to 11, characterised in that the variable derived from the machine parameter (PS, PW) is the cutting energy per volume (J / mm3) required to machine a predetermined volume of the allowance to be removed on the toothing or on the profile.

14. Method according to one of claims 1 to 13, characterised in that a characteristic value which characterises the grinding process for the ground workpiece is determined and output from the measured machine parameters (PS, PW) and / or from the variable (A) derived from the machine parameter (PS, PW).

15. Method according to one of claims 1 to 14, characterised in that the grinding tool is a grinding worm and the workpiece is a gear wheel.