Method for hard finishing workpiece having gear or profile on hard finishing machine tool

Abstract

The invention relates to a method for hard finishing a workpiece on a hard finishing machine tool, the hard finishing machine tool having a plurality of machine tool spindles, the workpiece being accommodated on a workpiece spindle and being machined by means of a tool accommodated on a tool spindle, the tool spindle being machined by engaging the tool with the workpiece. In order to be able to improve the monitoring of the state of a hard finishing machine tool, the invention provides that a test run is carried out on at least one machine tool axis at least during a period of a machining phase in which the tool does not engage with the workpiece, the machine tool axis being driven during the test run, a value being specified for at least one drive parameter, the invention relates to a method for determining at least one drive parameter for which a scalar value of the movement of the machine tool axis is expected, an actual value of the movement of the machine tool axis is detected, a comparison between the scalar value and the actual value is carried out by a monitoring device, and if the comparison between the scalar value and the actual value indicates that the difference between the scalar value and the actual value is outside an allowable range, the machine tool axis is determined. And the monitoring device outputs a signal.

Description

Technical Field

[0001] The present invention relates to a method for hard finishing workpieces having gears or contours on a hard finishing machine tool, wherein the hard finishing machine tool has a plurality of machine tool axes actuated under NC control to machine the workpiece. In this method, the workpiece is housed on a workpiece spindle and machined by means of a tool housed on a tool spindle, which is machined by engaging the tool with the workpiece. The workpiece machining phase begins with the introduction of the workpiece into the hard finishing machine tool and ends with the removal of the machined workpiece from the hard finishing machine tool. Background Technology

[0002] A general method is known from EP 3 168 700 A1. In this method, a workpiece is machined in a known manner under numerical control, wherein the tool engages with the workpiece to be machined, particularly with the workpiece to be ground. In addition to the actual machining time, various non-productive times occur, such as time for loading and unloading the workpiece and time for dressing the grinding tool.

[0003] Hard finishing machine tools wear down with operation, so timely maintenance and replacement of machine parts are essential. This also applies to situations involving unexpected events (e.g., collisions, overloads). It is crucial to ensure that no workpieces are produced when the machine is no longer in normal working order due to wear or damage.

[0004] Therefore, in practice, periodic benchmark runs of machine tools are typically performed to check their condition and, in particular, assess wear. During these benchmark runs, the machine tool axes (linear / rotary axes) are moved outside of actual operation, and their responses to specified control signals are recorded. This allows for checking whether the machine tool components involved are still in perfect condition or whether changes in the response to specified control signals due to wear have been detected. In the latter case, appropriate machine tool maintenance is required.

[0005] The performed cyclic benchmark runs provide a very accurate picture of the machine tool's condition. The precise results are obtained because the exact process moves can be performed under similar conditions during the benchmark runs, thus ensuring that disruptive variables have only a very small impact on the system.

[0006] However, a drawback is that it relies on the selected cycle for performing baseline runs, which determines the proximity of changes to the machine tool. In practice, due to the required time (since no workpieces can be machined during this period), baseline runs cannot be performed too frequently; however, this results in relatively long periods where the machine tool is not inspected (longer cycles between baseline runs). Longer baseline runs can also lead to problems with the machine tool's temperature behavior.

[0007] The fundamental problem of this invention is to develop a method of the type described above for improved condition monitoring of hard-finishing machine tools, and in particular their machine tool components, especially linear and rotary axes. The aim is to achieve the most closely meshed machine tool monitoring possible without sacrificing machine tool capacity. Furthermore, the aim is to more accurately detect the possible causes of wear. Summary of the Invention

[0008] The solution to this problem in this invention is characterized by performing a test run on at least one machine tool axis during at least a period of time in the machining phase when the tool is not engaged with the workpiece, wherein the machine tool axis is driven during the test run, wherein a value is specified for at least one drive parameter, for which a nominal value of movement of the machine tool axis is expected, wherein the actual value of movement of the machine tool axis is detected, wherein a comparison between the nominal value and the actual value is performed by a monitoring device, and wherein a signal is output by the monitoring device if the following condition occurs.

[0009] a) A comparison between the nominal value and the actual value indicates that the difference between the nominal value and the actual value is outside the permissible range, or

[0010] b) A comparison between the nominal and actual values ​​shows that the difference between the nominal and actual values ​​exceeds the allowable tolerance range compared to previously recorded differences.

[0011] In this regard, operating parameters are specified for at least one machine tool axis during the test run, and the axis's response to these parameters is recorded; if the deviation between the expected value (nominal value) and the measured value (actual value) exceeds the allowable range, a signal is issued (case a). However, trend analysis can also be performed in exactly the same manner, including values ​​recorded during previous test runs, to determine whether the machine tool axis will exceed the acceptable operating range (case b).

[0012] At least one time period of the machining phase is preferably within the time interval during which the workpiece is held on the workpiece spindle. However, other time periods of the machining cycle, such as periods for balancing or dressing, may also be considered.

[0013] Regarding the terminology used here, it should be noted that “test run” (especially as stated above) should be understood as monitoring of the machine tool axis during the routine machining of the workpiece, but this is only performed in machining sub-processes where the tool is not engaged with the workpiece, while “baseline run” is a separate test run of the (complete) machine tool axis, performed outside of the workpiece machining process.

[0014] Preferably, the test run is performed during the idle stroke, in which the tool moves relative to the workpiece without engaging it. Another preferred option is to perform the test run during acceleration or rotational acceleration and / or during machine axis braking. Of course, other parts of the machining cycle are also possible, such as rotating the rotary table or dressing.

[0015] Preferably, the test run is performed during the tool-grinding and / or the stage of aligning the workpiece in a hard finishing machine for machining, and / or the stage of performing a balancing operation, and / or the stage of performing a loading or unloading operation.

[0016] Another embodiment specifies that no signal is output before statistical analysis and / or another evaluation of multiple recorded values ​​are performed. This is to prevent signal output when there is only one "outlier" during the test run. The mentioned statistical evaluation is generally understood to refer to (further) analysis of the test data, which shows trends appearing in the recorded data.

[0017] The proposed method is particularly advantageous when using artificial intelligence (AI). Specifically, in this case, algorithms that take into account feedback (or empirical values) in the context of "machine learning" (ML) can be used to evaluate recorded values.

[0018] The detected values ​​are preferably accelerations recorded by an accelerometer, and / or currents used to drive machine tool axes, and / or control differences that occur when controlling machine tool axes, and / or signals that provide an indication of the deviation between nominal and actual values.

[0019] In the case of acceleration measurements, the onset of operational degradation can be inferred from corresponding low values ​​or from signals deviating from calculated or known levels within the time and / or frequency ranges. A significant increase in the current used to drive the machine tool axes during specified target movement (within the time and / or frequency ranges) can also be considered an indication of increased wear. This also applies to control errors that the controller must compensate for; if an increase in control error occurs here (within the time and / or frequency ranges), it can be concluded that the controller finds it more difficult to maintain the specified position due to wear. It should be noted that the evaluated and monitored signals can be viewed within both the time and frequency ranges.

[0020] If the assessment indicates that the value is outside the acceptable range, the output of the monitoring device signal can trigger a baseline run of the hard-finishing machine tool, in which all machine tool axes are actuated without the tool contacting the workpiece. Therefore, if an anomaly is detected during one or more test runs, this signal triggers the machine tool's baseline run.

[0021] The proposed method is preferably used for gear grinding.

[0022] The proposed concept can be used to determine the state of the machine tool, and in particular the state of machine tool components, during the machining cycle of a workpiece, thus enabling close and time-neutral monitoring. Therefore, time-consuming benchmark runs are performed only on a random sample basis, eliminating the need for cyclical execution and allowing them to be performed only when needed.

[0023] Close time monitoring improves monitoring quality and advantageously reduces non-productive time. This results in a reduction in the number of workpieces that are not properly processed (due to wear or defective machine tool parts). Furthermore, it greatly simplifies the analysis of defect causes and provides favorable conditions for preventative machine tool maintenance (the concept of "predictive maintenance").

[0024] Another advantage is that it makes it easier to identify the cause, especially the cause of sudden wear and tear, and the cause can be identified more precisely.

[0025] Another advantage is that the availability of machine tools for efficient machining of workpieces is increased by conducting test runs during available non-production time, thus reducing or avoiding unplanned machine tool downtime by utilizing the described methods.

[0026] Finally, it can advantageously avoid changes in the critical temperature of the machine tool. Attached Figure Description

[0027] The accompanying drawings illustrate embodiments of the present invention.

[0028] Figure 1 The wear process of components in a gear grinding machine is schematically illustrated, with the wear index plotted over time.

[0029] Figure 2 The wear process of machine tool components is schematically illustrated, with wear indices plotted across multiple measurements.

[0030] Figure 3 A flowchart illustrating how to determine the criteria used to perform baseline operations on a machine tool is shown.

[0031] Figure 4 The concept of data exchange provided according to an embodiment of the proposed method is illustrated schematically.

[0032] Figure 5 The diagram schematically illustrates the changes of the three machine tool axes over time, with rectangles marking the areas where test runs are planned.

[0033] Figure 6 A flowchart illustrating a possible embodiment for implementing the proposed concept is shown schematically. Detailed Implementation

[0034] The following explanation is based on the described concept, which involves performing a test run on at least one machine tool axis during a machining phase in which the tool does not contact the workpiece, particularly during gear grinding. During this test run, the machine tool axis is actuated according to at least one predefined operating parameter, and the axis's response to the operating parameter is recorded. The recorded values ​​are evaluated by a monitoring device. If the evaluation indicates that the value is outside the allowable range, the monitoring device outputs a signal. In this case, irregularities exist in the machine tool, which are specifically attributable to corresponding wear. Therefore, a benchmark run is subsequently performed, in which all axes of the machine tool are operated and inspected according to a defined program.

[0035] Compared to previously known procedures, the machine tool condition is now assessed using all or at least some of the idling strokes and acceleration phases of the axes. In principle, all movements during the machining process (especially tool dressing, workpiece alignment, etc.) can be used for this, where the workpiece and tool do not come into contact.

[0036] If an abnormal signal occurs during the idling stroke, a reference run can be automatically initiated (to determine the precise value), or a message indicating that a reference run should be performed can be output to the control system. Additionally, the machine tool can be stopped to prevent indirect damage.

[0037] This means that additional information about the machine tool's status is recorded in a time-neutral manner, enabling virtually complete monitoring and allowing for very rapid initiation of corresponding measures after an anomaly is detected; thus, it can prevent the production of scrap.

[0038] In addition, a comprehensive preventative machine tool maintenance concept (“preventative maintenance”) can be installed.

[0039] This is particularly advantageous in cases of sudden wear, allowing for the rapid identification of components, such as machine tool parts, that require replacement. If necessary, the optimal time for measurement can be predicted here by extrapolating characteristic values.

[0040] During the workpiece machining phase where the tool is not engaged with the workpiece, especially when dressing the tool, balancing, rotating the turntable, or during other machine tool movements required in the machining cycle, record data from each axis.

[0041] These idle strokes (i.e., the machine tool movement during which the tool is not engaged with the workpiece) can be specifically idling strokes during fine or rough dressing (Y-axis), feed strokes during dressing and grinding (X-axis), idle strokes during grinding (Z-axis), or near reference positions used for balancing or rotation. Furthermore, acceleration or deceleration phases of the machine tool axes must be considered. Similarly, these are ideally suited for stages of machining processes involving balancing, rotation, workpiece alignment, turret (rotary table) dressing, and workpiece changeover.

[0042] The recorded measurement signals can be signals from accelerometers mounted on machine tool components, internal control signals (such as current or control differential), and other signals from existing sensors (such as current clamps and encoders).

[0043] Eigenvalues ​​can be calculated to analyze signals (e.g., wear index). Signal curves can also be used and evaluated using neural networks.

[0044] Signal processing is preferably performed on an industrial PC.

[0045] Signal curves, evaluations, and analyses can be stored in a local database on the PC. Additionally, there are options for storing and evaluating data in the cloud.

[0046] The proposed concepts can be applied to specific components, machine tools, or parts, or to multiple machine tools. Methods across different machine tool types are envisioned.

[0047] Regardless of the scope of the concept proposed, there is always an idle stroke for all axes involved in machining. This can occur during process preparation, during actual grinding, during dressing, or upon process completion.

[0048] Individual axes (especially the Y-axis) are particularly important for the quality of the grinding process. Monitoring of this axis can be performed as follows: During dressing, standard dressing techniques involve no-load strokes during fine dressing (two or more dressing strokes – vibration stops) or no-load strokes during rough dressing (an odd number of dressing strokes). During the no-load strokes, the axis moves at multiple defined speeds across almost the entire transverse range. Similarly, rotary axes (e.g., during centrifugal processes after dressing) may be relevant. The described test runs can then be performed on all these axes. Of course, other no-load strokes in other parts of the machining process also help to check various machine tool components, equivalent to the dressing process.

[0049] Figure 1 The wear index IND as a function of time t is shown, which generates a corresponding reference index IR during the baseline operation; the determined reference index IR is marked with a cross. Regarding the wear index, it should be noted that this is a value that can be calculated from various values ​​or signals related to or typical of wear.

[0050] For example, the time between two benchmark runs is one week. The wear index must be below the warning limit W, which is considered safe. If it is still above the error limit F, it is expected that it will no longer be possible to produce the workpiece correctly.

[0051] During the idling stroke IL, wear indices are recorded at much shorter time intervals, which are entered only as examples of the time between the last two reference indices IR; the corresponding wear indices determined during the idling stroke IL are marked with dots. A large number of wear indices are determined during the idling stroke IL compared to the reference indices IR, which are determined only over relatively long time intervals; the latter are preferably determined continuously during the machining process.

[0052] It can be seen that there is a sudden increase in wear between the penultimate reference index IR and the last reference index IR, because the penultimate reference index is still well below the warning limit W, but the next reference index is already above the error limit F. Therefore, the wear occurring on the machine tool axis involved here can only be detected when multiple workpieces have been produced and are therefore about to be scrapped. However, the proposed method and a larger number of wear indices during the idling stroke can be used to determine a relatively sharp rise in the wear index IND, allowing maintenance measures to be taken before the warning limit W is reached.

[0053] Figure 2 This illustrates the recommendation for statistical or trend-based assessment when monitoring the wear index IND. It shows how the magnitude of the wear index IND is represented by the number of measurements taken on three different machine tool axes or their response to a given operating parameter (represented by dots, crosses, or squares). It can be seen that during the idling stroke IL, for axes marked with dots and crosses, the machine tool axes or their response to a given operating parameter are not significant and exhibit low levels of wear index IND. However, it can be seen that the wear index IND of the axes marked with squares gradually increases, thus showing a baseline trend TREF.

[0054] In statistical analysis, outliers (marked as AUS) may be disregarded.

[0055] Meanwhile, the wear index IND of the shaft marked with dots indicates that HAUF accumulation occurred when the wear index value increased significantly starting from the 64th measurement, indicating a sudden and significant increase in wear on this machine tool shaft.

[0056] Therefore, in the current situation, it is important to consider that the algorithm used only initiates benchmark runs under certain circumstances, or indicates that benchmark runs should be performed. Due to various influencing factors, the information values ​​of the signals from the test run are significantly lower than those from the benchmark run. In this regard, benchmark runs are not always performed if limits are exceeded during the test run, but only when a specific accumulation of an anomaly is detected or a corresponding trend is identifiable. The storage algorithm used to select whether to perform a benchmark run is preferably configurable and / or self-learning.

[0057] Figure 3The standard for optimizing the execution of a baseline run is illustrated. As explained, operating parameters are continuously analyzed during the idle stroke of the machine tool axis, and these parameter values ​​are checked to see if they are within specified limits (“compliant?”). If the parameter values ​​are not within the specified limits, a baseline run is executed, and the system responds accordingly.

[0058] The datasets available from the idling stroke, as well as the precise (highly accurate) data available from benchmark runs, are influenced by the concept of "machine learning" as part of the algorithm, where an optimized training dataset is generated through continuous labeling, which can then be used to examine the data obtained during the idling stroke.

[0059] Therefore, another aspect of the proposed method is the use of a self-learning algorithm that optimizes the correlation between the analysis of the test run and the results of the benchmark run after each benchmark run.

[0060] Figure 4 The data processing concept of a possible implementation of the proposed method is schematically illustrated. The left side of the figure shows a schematic diagram of a machine tool undergoing workpiece grinding, preheating, dressing, and alignment. Accelerometers are used to determine whether the machine tool is in normal operating condition during each process. Corresponding displays can be provided via a human-machine interface (HMI).

[0061] The machine tool control system controls the various processes and transmits drive and / or vibration data to an additional PC via digital-to-analog converter output (DAU), trajectory, sensor, or other data from the process.

[0062] The PC contains software with appropriate algorithms for evaluating the received data. For this purpose, values ​​recorded by the accelerometer and / or analog measurement outputs are transmitted to measurement inputs available to the software. The software also connects to a database storing the process data.

[0063] PCs can connect to the cloud.

[0064] Figure 5 The movement of the three axes and their changes over time are schematically illustrated. Rectangles mark the time periods during which test runs can be performed in areas where there is no engagement between the tool and the workpiece.

[0065] The relevant signals measured here are the power consumption of the axis (in this case, the Y-axis of the machine tool), the control differential in the controller, and the acceleration measured by the accelerometer. These parameters are typically present in the machine tool itself when the measurements are recorded to monitor the dressing process.

[0066] The idling stroke can be identified using internal control signals (dressing stroke counter, dressing stage, etc.) and a specified position of the shaft (via an integrated scale); the corresponding signals can be "cut off" ( Figure 5 (The area marked with a rectangle in the middle).

[0067] During constant operation, the signal is evaluated by calculating statistical characteristic values, but during acceleration and deceleration phases, the signal is also evaluated using machine learning algorithms. Choosing machine learning methods to evaluate continuous operation can also be advantageous.

[0068] The calculated wear value (see) Figure 1 and Figure 2 The wear index (IND) is stored in the database and compared with previously defined limits and / or taught limits so that a benchmark run is triggered when it is exceeded.

[0069] Therefore, its limit values ​​(with automatic teaching) are determined by the previous idling stroke and reference run.

[0070] It should be noted that, as already mentioned, the signal quality during the idling stroke in a test run is typically worse than that of a baseline run. Therefore, the algorithm is preferably used to evaluate whether a baseline run is necessary, particularly by means of statistical methods or by using the concept of "machine learning" (especially by means of similarity analysis, which can induce modifications to the values).

[0071] In this respect, it is unnecessary to reduce to purely static values ​​or impose restrictions.

[0072] As mentioned earlier, if a baseline run is triggered, the height accuracy of the baseline run can be correlated with the results of the idling stroke, thereby optimizing the dynamic limits.

[0073] Figure 6 A flowchart illustrating this method is shown.

Claims

1. A method for hard finishing a workpiece having gears or contours on a hard finishing machine tool, wherein the hard finishing machine tool has a plurality of machine tool axes actuated under NC control to machine the workpiece, wherein the workpiece is received on a workpiece spindle and machined by means of a tool received on a tool spindle, the tool spindle being machined by engaging the tool with the workpiece, wherein the workpiece machining phase begins with introducing the workpiece into the hard finishing machine tool and ends with removing the machined workpiece from the hard finishing machine tool. Its features are, A test run shall be performed on at least one machine tool axis during at least one time period of the machining phase in which the tool is not engaged with the workpiece. The machine tool axis is driven during the test run, and a value is specified for at least one drive parameter, representing the nominal value of the expected movement of the machine tool axis for that at least one drive parameter. The actual value of the movement of the machine tool axis is detected. The comparison between the nominal value and the actual value is performed by a monitoring device, and the monitoring device outputs a signal if the following conditions are met. a) The comparison between the nominal value and the actual value indicates that the difference between the nominal value and the actual value is outside the permissible range, or b) The comparison between the nominal value and the actual value indicates that the difference between the nominal value and the actual value has exceeded the allowable tolerance range compared to the previously recorded difference value.

2. The method according to claim 1, characterized in that, At least one time period of the processing stage is within the time interval during which the workpiece is accommodated on the workpiece spindle.

3. The method according to claim 1, characterized in that, The test run is performed during an idle stroke, in which the tool moves relative to the workpiece without engaging it.

4. The method according to claim 1, characterized in that, The test run is performed during acceleration or rotational acceleration and / or during machine tool axis braking.

5. The method according to claim 1, characterized in that, The test run is performed during the dressing phase of dressing the tool, and / or the phase of aligning the workpiece in the hard finishing machine for machining, and / or the phase of performing a balancing operation, and / or the phase of performing a loading or unloading operation.

6. The method according to claim 1, characterized in that, The signal is not output until a statistical evaluation is performed on multiple detection values.

7. The method according to claim 1, characterized in that, The detection values ​​are evaluated using an algorithm that considers empirical values ​​in the context of machine learning.

8. The method according to claim 1, characterized in that, The detected value is an acceleration detected by an accelerometer, and / or a current driving the machine tool axis, and / or a control difference occurring during the control of the machine tool axis, and / or a signal providing an indication of the deviation between the nominal value and the actual value.

9. The method according to claim 1, characterized in that, If the assessment indicates that the value is outside the permissible range, the monitoring device outputs the signal to perform a baseline operation of the hard finishing machine tool, wherein all machine tool axes are actuated without the tool engaging the workpiece.

10. The method according to claim 1, characterized in that, The method described is a gear grinding method.

Images