Automatic process control system and method in a gear manufacturing machine

Abstract

A method for monitoring a machining process is proposed, in which a tooth surface of a workpiece (23) of a preformed tooth is machined by means of a finishing machine (1). During the method, a plurality of measured values are detected during the machining engagement of the finishing tool (16) with the workpiece. Among the measured values there is a value of a power indicator, which shows the instantaneous power consumption of the tool spindle during machining of the tooth surface of the workpiece. A normalization operation is applied to the values of at least some of the measured values or variables derived from the measured values in order to obtain normalized values. The normalization operation is related to at least one process parameter: a geometric parameter of the finishing tool, in particular its outer diameter; a geometric parameter of the workpiece; and a set parameter of the finishing machine, in particular the radial feed and the axial feed.

Description

TECHNICAL FIELD

[0001] The present invention relates to a method for monitoring a finishing machine for tooth production, in particular a roller finishing machine. Furthermore, the invention also relates to a finishing machine set up for carrying out such a method, a computer program for carrying out such a method and a computer readable medium having such a computer program. BACKGROUND

[0002] Hard finishing of pre-finished toothings is a very demanding method in which even the smallest deviation from the process specification can result in the finished workpiece having to be regarded as scrap ("NIO parts", where NIO stands for "not in order"). This problem can be explained particularly well by means of continuous roller grinding, but also applies to other roller-based finishing methods, such as partial gear grinding, gear honing or hard gear shaving. However, similar problems also occur in non-roller methods, such as discontinuous or continuous profile grinding, at smaller dimensions.

[0003] In the case of continuous gear grinding, a pre-finished gear blank is machined in roller engagement with a worm-shaped profiled grinding wheel (grinding worm). Gear grinding is a very demanding generating machining method which is based on a large number of simultaneous, high-precision individual movements and is influenced by a large number of boundary conditions. Information on the basic principles of continuous gear grinding can be found, for example, in the book "Continuous Generating Gear Grinding" by H. Schriefer et al., Reishauer AG self-published, Wallisellen 2010, ISBN 978-3-033-02535-6, Chapter 2.3 ("Basic Methods of Generating Grinding"), pages 121 to 129.

[0004] In theory, the tooth surface shape in continuous gear grinding is determined only by the modified profile shape of the worm and the set data of the machine. In practice, however, deviations from the ideal state occur in automated production, which can have a decisive influence on the grinding result.

[0005] Conventionally, the quality of the toothings produced in the gear grinding method is evaluated after the end of the machining by toothing measurements outside the machining machine ("off-line") according to a large number of measurement variables. Here, different standards exist which specify how the toothings are to be measured and how it is to be checked whether the measurement results are within or outside the tolerance specifications. A summary on such toothing measurements can be found, for example, in the aforementioned work by Schriefer et al., Chapter 3 ("Quality Assurance in Continuous Generating Gear Grinding") on pages 155 to 200.

[0006] It is known from the prior art that corrections are made on the machine on the basis of toothing measurements in order to eliminate the machining errors identified. This is discussed in the aforementioned work by Schriefer et al. in Chapter 6.10 ("Analysing and Eliminating Gear Tooth Deviations") on pages 542 to 551.

[0007] However, because only a sample test is usually carried out for time and cost reasons at the toothing check, machining errors are often identified very late. This can lead, in some cases, to a considerable proportion of a production batch having to be discarded as NIO parts. It is therefore desirable that, ideally, machining errors are identified as early as possible "on-line" during the machining before the machining errors reach such a degree that the workpieces have to be discarded as NIO parts.

[0008] To this end, it is desirable to provide an automated process monitoring which enables the identification of undesirable process deviations, from which indications about possible machining errors are obtained, and targeted changes to the settings of the machine so that machining errors are avoided or at least reduced.

[0009] If machining errors are identified only later, for example at the EOL test (EOL = End of Line), the process monitoring should ideally also allow conclusions to be drawn about the process deviations ex post.

[0010] In the prior art, suitable strategies for automated process monitoring during tooth machining are known only to a certain extent.

[0011] It is therefore known from DE 10 2014 015 587 A1 that parameters on a gear cutting machine are monitored and a tooth check is only carried out when a specific measured machine parameter deviates from the desired value.

[0012] Therefore, the company report "NORDMANN Tool Monitoring", version of October 5, 2017, was taken from https: / / www.nordmann.eu / pdf / praesentation / Nordmann_presentation ENG.pdf The call describes different measures for tool monitoring at general-purpose cutting machines (page 3). In the report, application examples in different cutting machining methods are shown, wherein a brief example of a small number of methods relevant for gear machining is also included, in particular gear hobbing (pages 41 and 42), hard skiving (page 59) and honing (page 60). Continuous gear grinding is only mentioned briefly (e.g. page 3 and page 61).

[0013] Methods for automated process monitoring in various different machining methods are also known, inter alia, from US 5,070,655, US 3,809,970, US 4,894,644 and Klaus Nordmann, "Prozessüberwachung beim Schleifen und Abrichten", Schleifen+Polieren 05 / 2004, Fachverlag Velbert (Germany), pages 52-56. However, the finishing of toothings is not discussed in detail here either.

[0014] One difficulty in process monitoring when machining toothings is the fact that the monitored measurement variables are related in a highly complex manner to a large number of geometric properties of the tool (in the case of grinding worm gears, for example, the diameter, the module, the number of threads, the pitch angle, etc.), to geometric properties of the workpiece (for example, the module, the number of teeth, the helix angle, etc.) and to set parameters on the machine (for example, the radial infeed, the axial feed, the rotational speeds of the tool and workpiece spindles, etc.). Due to this diverse, complex correlation, it is extremely challenging to draw direct conclusions from the monitored measurement variables about specific process deviations and the resulting machining errors. It is also extremely difficult to compare the measurement variables in different machining processes with one another. Additional challenges arise when using reconditionable tools. By reconditioning, the diameter of the tool, and thus also the machining conditions, are changed in the course of machining a series of workpieces. Therefore, even if all other framework conditions remain the same, even within the same series of workpieces, the monitored measurement variables in different reconditioning cycles cannot be directly compared with one another.

[0015] The known methods for process monitoring have so far not taken sufficient account of these properties of tooth machining. SUMMARY

[0016] It is an object of the present application to propose a method for process monitoring during tooth machining, which enables process deviations to be identified and eliminated in a targeted manner. The method should also be suitable in particular for use with reworkable tools.

[0017] The object is achieved by the method according to the application. Further embodiments are given in the following description.

[0018] A method for monitoring a machining process is given, in which a tooth surface of a workpiece with a preformed tooth is machined in a finishing machine. The finishing machine has a tool spindle for driving a finishing tool in rotation about a tool axis and a workpiece spindle for driving the workpiece with the preformed tooth in rotation. The method comprises:

[0019] detecting a plurality of measurement values during the finishing tool-workpiece machining engagement; and

[0020] applying a normalization operation to at least some of the measurement values or to values derived from the measurement values in order to obtain normalized values,

[0021] wherein the normalization operation is related to at least one process parameter, wherein the at least one process parameter is selected from the group consisting of a geometric parameter of the finishing tool, a geometric parameter of the workpiece and a setting parameter of the finishing machine.

[0022] It is thus proposed to detect measurement values on the finishing machine and to subject at least some of these measurement values or values derived therefrom to a normalization operation. The normalization operation takes into account the influence of one or more process parameters on the measurement values, in particular a geometric parameter of the finishing tool (in particular its dimensions, in particular its outer diameter), a geometric parameter of the workpiece and / or a setting parameter of the finishing machine (in particular the radial infeed, the axial feed and the rotational speeds of the tool and workpiece spindles). The resulting normalized values are thus independent or at least significantly less dependent on the mentioned process parameters. Due to the normalization operation, the normalized values are comparable between different machining processes, even if these process parameters differ.

[0023] The normalization operation is then particularly important when the measurement values comprise values of a power indicator, which shows the instantaneous power consumption of the tool spindle during machining of the tooth surface of the workpiece. The detected power indicator can in particular be a measure of the power consumption of the tool spindle. This power indicator is in particular influenced by the mentioned parameters. It is therefore particularly advantageous to apply the normalization operation to the values of the power indicator or to variables derived therefrom.

[0024] The normalization operation is preferably based here on a model which describes the expected correlation of the measured values with the mentioned parameters. If the measured values are values of a power indicator, the model preferably describes the correlation of the process power, i.e. the mechanical or electrical power required for the machining process performed, with the mentioned parameters. The model of the process power can be based in particular on a force model which describes the expected correlation of the cutting forces acting at the location of the contact between the finishing tool and the workpiece with the geometric parameters of the finishing tool, the geometric parameters of the workpiece and the setting parameters of the finishing machine. The model of the process power can also take into account the length of the lever arm acting between the tool axis and the contact point located between the finishing tool and the workpiece. The lever arm length can be approximated in particular by the outer diameter of the finishing tool. Furthermore, the model of the process power can take into account the rotational speed of the tool spindle.

[0025] The normalization operation can comprise for example multiplying the detected measured values or variables derived therefrom by a normalization factor. But also more complex normalization operations are conceivable. If the measured values comprise values of a power indicator, the normalization factor can be in particular an inverse power variable calculated from the model of the process power for the machining situation prevailing, or a variable derived therefrom.

[0026] The normalization operation is preferably applied directly to the detected measured values, if necessary after filtering. The normalization operation is advantageously carried out in real time, i.e. also during the machining process, in particular also during the machining of the respective workpiece, i.e. also during the machining engagement of the finishing tool with the workpiece. Thereby, the normalized values are directly available during the machining process and can be used in real time for monitoring the machining process.

[0027] It is particularly advantageous to analyze the normalized values in real time in order to determine unallowable process deviations also during the machining process. This makes it possible to identify workpieces for which an unallowable process deviation was determined immediately after their machining and, if necessary, to remove said workpieces from the workpiece batch in order to subject them to individual treatment. For example, such workpieces can be subjected to individual measurements or directly classified as NIO components.

[0028] In an advantageous embodiment form, the method comprises calculating characteristic variables of the machining process from the measured values or from variables derived therefrom. Calculating characteristic variables from the measured values or from variables derived therefrom is advantageous regardless of whether a normalization operation is carried out in the process of the method. In some embodiment forms, at least one of the characteristic variables is a normalized characteristic variable, i.e. a normalization operation is applied at any time in the process of calculating the characteristic variable. This can be achieved by calculating the characteristic variables from the normalized measured values. But this can also be achieved by first applying a normalization operation to intermediate results, i.e. to variables derived from the measured values, when calculating the characteristic variables.

[0029] At least one of the characteristic variables is preferably specific to the machining process. Thus, it is preferred that not only statistical variables such as mean value, standard deviation, etc. which can be formed independently of the specific machining process are involved, but also characteristic variables which take into account the characteristics of the specific machining process.

[0030] Preferably, at least one of the characteristic variables is related to a predefined machining error of the workpiece. It is particularly advantageous if there is a one-to-one relationship, in particular a simple proportional relationship, between the characteristic variable and the variable of the machining error. It is thereby possible to obtain a direct indication of the production of a specific machining error by monitoring the characteristic variables of different workpieces. This enables even inexperienced operators to correctly interpret the characteristic variables and intervene in a corrective manner.

[0031] Particular advantages arise when these characteristic variables are directly related to the results in the gear measurement. To this end, the method can comprise:

[0032] performing a gear measurement on the selected workpieces in order to determine for each of the workpieces at least one gear measurement variable characterizing a predefined machining error; and

[0033] determining a correlation parameter characterizing a correlation of at least one characteristic variable and at least one gear measurement variable.

[0034] The calculation of at least one of the characteristic variables can in particular comprise a spectral analysis of the measurement values, in particular values of a (preferably normalized) power indicator and / or values of an acceleration sensor. It is particularly preferred here to evaluate spectral components at multiples of the tool rotational speed and / or the workpiece rotational speed. The calculation of the respective characteristic variable is thereby specific to the machining process. Such a spectral analysis is also advantageous if no normalization is performed, as is the case, for example, in the case of values of an acceleration sensor.

[0035] If the finishing process is a rolling process in which a finishing tool and a workpiece roll engage, in particular a gear grinding process, it is advantageous if the characteristic variables comprise at least one of the following variables:

[0036] a cumulative pitch indicator which is calculated at the rotational speed of the workpiece spindle by evaluating spectral components of the measurement values of the power indicator, in particular normalized, and which is associated with a cumulative pitch error of the workpiece,

[0037] a profile shape indicator which is calculated at the gear engagement frequency by evaluating spectral components of the measurement values of the acceleration sensor, in particular, and which is associated with a profile shape deviation of the workpiece.

[0038] It is also advantageous for the characteristic variable to include a wear indicator, wherein the wear indicator is calculated from a low-pass filtered spectral component of a measurement of the power indicator, in particular a normalized power indicator, and is related to the degree of wear of the finishing tool.

[0039] Advantageously, a process deviation of the machining process from the desired process is determined from the course of change of at least one of the characteristic variables of the workpieces. For this purpose, the value of the selected characteristic variable is advantageously related to the value of another characteristic variable or another process variable. The machining process is then advantageously adjusted in order to reduce the process deviation, or the limit values used in the above-mentioned real-time analysis are adjusted in order to recognize an impermissible process deviation. The determination of the process deviation can here be carried out by means of a trained machine learning algorithm.

[0040] The method can include storing data sets in a database, wherein the data sets include an unambiguous designation of a workpiece, at least one process parameter and at least one of the characteristic variables. Furthermore, the method can also include:

[0041] calling up data sets of a plurality of workpieces from a database; and

[0042] graphically outputting values of at least one of the characteristic variables of the plurality of workpieces or values derived therefrom.

[0043] The steps can be carried out, for example, in a web browser, since the computing and storage space requirements of these steps are very moderate compared to the processing of raw data.

[0044] The method preferably provides a recalculation of the normalization only when at least one of the process parameters is changed. The recalculation of the normalization then preferably includes the application of the mentioned model with the changed process parameter.

[0045] The recalculation of the normalization can be carried out, in particular, on the basis of a changed dimension of the finishing tool, in particular its outer diameter, and can include a compensation in respect of the changed dimension. This is particularly important in the case where the dimension is variable during the machining of a series of workpieces, as is often the case with reconditionable tools. In this way, the characteristic variables determined in different reconditioning cycles can be directly compared with one another.

[0046] The application also provides a finishing machine for machining the tooth surface of a workpiece of a preformed tooth. The finishing machine has a tool spindle for rotationally driving a finishing tool about a tool axis, a workpiece spindle for rotationally driving a workpiece of a preformed tooth, a control device for controlling the machining process of the workpiece by means of the finishing tool, and a process monitoring device. The process monitoring device is specifically set up to carry out the mentioned method.

[0047] To this end, the process monitoring device preferably has:

[0048] a detection device for detecting a plurality of measurement values during the machining engagement of the finishing tool with the workpiece; and

[0049] a normalization device for applying a normalization operation to at least a portion of the measurement values or to values derived from the measurement values in order to obtain normalized values,

[0050] wherein the normalization operation is related to at least one process parameter, wherein the at least one process parameter is selected from the group consisting of a geometric parameter of the finishing tool, a geometric parameter of the workpiece, and a set parameter of the finishing machine.

[0051] Preferably, the normalization device is configured to perform the normalization operation in real time during the machining engagement of the finishing tool with the workpiece.

[0052] In some embodiments, the process monitoring device has an error recognition device configured to analyze the normalized values in real time in order to determine an inadmissible process deviation.

[0053] The finishing machine can have a workpiece handling device configured to automatically remove a workpiece for which an inadmissible process deviation has been determined.

[0054] In some embodiments, the process monitoring device has a characteristic variable calculation device for calculating characteristic variables of the machining process from the measurement values or from values derived therefrom. The characteristic variable calculation device can be configured to perform a spectral analysis of the measurement values, values derived therefrom, or normalized values for at least one of the characteristic variables and, in this regard, to evaluate, in particular, spectral components at multiples of the rotational speed of the tool spindle and / or the workpiece spindle.

[0055] The process monitoring device can have a data communication device for transmitting a data set to a database, wherein the data set includes an unambiguous designation of the workpiece, at least one process parameter, and at least one of the characteristic variables.

[0056] The process monitoring device can have a deviation determination device for determining a process deviation of the machining process from a desired process from values of at least one of the characteristic variables of a plurality of workpieces. The deviation determination device can include a processor device programmed to execute a trained machine learning algorithm in order to determine the process deviation.

[0057] The process monitoring device can have a normalization calculation device for recalculating the normalization operation when at least one of the process parameters changes. The normalization calculation device is preferably configured to apply a model describing the expected correlation of the measurement values with the process parameters, in particular a model of the process force or the process power, when recalculating the normalization operation. The normalization calculation device can advantageously be configured to perform a compensation with respect to the dimensions of the finishing tool, in particular with respect to its outer diameter.

[0058] The data detection device, the normalization device, the error recognition device, the characteristic variable calculation device, the normalization calculation device, the deviation determination device and the data communication device can at least partially be implemented in software for execution on one or more processors of the process monitoring device.

[0059] The present application also provides a computer program. The computer program comprises instructions which cause the process monitoring device, in particular one or more processors of the process monitoring device, in a finishing machine of the above-mentioned type to perform the above-mentioned method. The computer program can be stored in a suitable storage device, for example in a computer which is configured separately from the machine control device.

[0060] Furthermore, the present application provides a computer-readable medium on which the computer program is stored. The medium can be a non-volatile medium, for example a flash memory, a CD, a hard disk, etc. BRIEF DESCRIPTION OF DRAWINGS

[0061] A preferred embodiment form of the present application is described below with reference to the accompanying drawings, which are used for illustration only and are not to be construed as being limiting. In the drawings:

[0062] Fig. 1 A schematic view of a gear grinding machine is shown;

[0063] Fig. 2 An enlarged detail in region II in Fig. 1 is shown;

[0064] Fig. 3 An enlarged detail in region III in Fig. 1 is shown;

[0065] Fig. 4 A diagram showing two exemplary variation curves of the current consumption of the tool spindle during machining of a tooth surface is shown;

[0066] Fig. 5 A diagram showing the maximum values of the current consumption of the tool spindle for a plurality of workpieces as a function of the ground worm outer diameter is shown; squares: non-normalized values; crosses: normalized values.

[0067] Fig. 6a spectrum of the current consumption of the tool spindle during machining of the workpiece is shown;

[0068] Fig. 7 a spectrum of the measurement values of the acceleration sensor during machining of the workpiece is shown;

[0069] Fig. 8 a diagram is shown which shows the values of the cumulative pitch indicator of a plurality of workpieces as a function of the grinding worm position;

[0070] Fig. 9 a diagram is shown which shows the values of the profile shape indicator of a plurality of workpieces as a function of the grinding worm position; triangle: first workpiece spindle; cross: second workpiece spindle;

[0071] Fig. 10 a diagram is shown which shows the values of the wear indicator of a plurality of workpieces as a function of the grinding worm position;

[0072] Fig. 11 a diagram is shown which shows the values of the vibration indicator of a plurality of workpieces as a function of the grinding worm position;

[0073] Fig. 12 a flow chart of a method for monitoring the machining of a workpiece batch is shown;

[0074] Fig. 13 a flow chart of a method for automatically identifying and correcting process deviations is shown;

[0075] Fig. 14 a flow chart of a method for graphically outputting characteristic variables and information derived therefrom is shown; and

[0076] Fig. 15 a schematic block diagram of the functional units of a process monitoring device is shown. DETAILED DESCRIPTION

[0077] Exemplary structure of a gear grinding machine

[0078] In Fig. 1The image exemplifies a gear grinding machine 1 as a finishing machine for machining the tooth surface of a workpiece with pre-formed teeth. The machine has a machine tool 11 on which a tool holder 12 is displaceably guided along the radial feed direction X. The tool holder 12 carries an axial slide 13, which is movable relative to the tool holder 12 along the feed direction Z. A grinding head 14 is mounted on the axial slide 13 and is pivotable about a pivot axis extending parallel to the X-axis (the so-called A-axis) to adapt to the helix angle of the tooth to be machined. The grinding head 14, in turn, carries a movable slide on which a tool spindle 15 is displaceable relative to the grinding head 14 along a movement axis Y. A worm-shaped grinding wheel (grinding worm) 16 is clamped on the tool spindle 15. The grinding worm 16 is driven by the tool spindle 15 to rotate about the tool axis B.

[0079] Machine tool 11 also carries a pivotable tool post 20 in the form of a turret, which is capable of pivoting about axis C3 between at least three positions. Two identical workpiece spindles are mounted on the tool post 20, diagonally opposite each other. Fig. 1 Only one workpiece spindle 21 and its associated tailstock 22 are visible. Fig. 1 The workpiece spindle, visible in the image, is in a machining position, in which the workpiece 23 clamped thereon can be machined by means of a grinding worm gear 16. Another spindle, offset by 180°, and... Fig. 1 The invisible workpiece spindle is in the workpiece changing position, where the machined workpiece can be removed from the spindle and a new blank can be clamped. A dressing device 30 is installed at a 90° offset from the workpiece spindle.

[0080] All driven axes of the gear grinding machine 1 are digitally controlled by a machine control unit 40. The machine control unit 40 includes multiple axis modules 41, a control computer 42, and an operation panel 43. The control computer 42 receives operation commands from the operation panel 43 and sensor signals from various sensors of the gear grinding machine 1, and calculates control commands for the axis modules 1 from these signals. The control computer also outputs operating parameters to the operation panel 43 for display. The axis modules 41 provide control signals at their output terminals to the respective machine axes (i.e., at least one actuator for driving the associated machine axis, such as a servo motor).

[0081] The process monitoring device 44 is connected to the control computer 42. The process monitoring device continuously receives multiple measurement values ​​from the control computer 42 and, if necessary, from other sensors. On one hand, the process monitoring device 44 continuously analyzes the measurement values ​​to identify machining errors early and remove the relevant workpieces from the machining process. On the other hand, the process monitoring device 44 calculates different characteristic variables of the finally machined workpiece from the measurement values. These processes will be described in more detail below.

[0082] The process monitoring device 44 transmits a dataset to the database server 46 for each workpiece. The dataset contains a clear workpiece identifier along with selected process parameters and characteristic variables. The database server 46 stores these datasets in a database, enabling subsequent retrieval of the relevant dataset for each workpiece. The database server 46, with its database, can be located inside the machine or remotely. The database server 46 can connect to the process monitoring device 44 via a network, as is the case in [the context of the process monitoring device 44]. Fig. 1 As represented by the cloud. In particular, the database server 46 can be connected to the process monitoring device 44 via a LAN within the machine or enterprise, via a WAN, or via the Internet.

[0083] Client 48 can connect to database server 46 to retrieve, receive, and evaluate data from database server 46. This connection can also be made via a network, particularly via a LAN, WAN, or the Internet. Client 48 may include a web browser, which allows visualization of the received data and its evaluation. The client does not need to meet special requirements for computing performance. The client application also does not require large network bandwidth.

[0084] exist Fig. 2 Enlarged to show Fig. 1 Part II of the image shows the tool spindle 15 and the grinding worm 16 clamped thereon. A detector 17 is pivotally mounted on a fixed portion of the tool spindle 15. The detector 17 is capable of selectively detecting... Fig. 2 The detector 17 pivots between a measuring position and a resting position. In the measuring position, the detector 17 can be used to tactilely measure the teeth of the workpiece 23 on the workpiece spindle 21. This occurs "online," meaning while the workpiece 23 is still on the workpiece spindle 21. This allows for early detection of machining errors. In the resting position, the detector 17 is located in an area where it is protected from collisions with the workpiece spindle 21, tailstock 22, workpiece 23, and other components on the workpiece carrier 20. The detector 17 is in this resting position during workpiece machining.

[0085] A centering probe 24 is arranged on the side of the workpiece 23 facing away from the grinding worm 16. In the present example, the centering probe 24 is configured and arranged in accordance with the document WO 2017 / 194251 Al. Reference is made to the mentioned document for details on the mode of operation and arrangement of the centering probe. In particular, the centering probe 24 can comprise an inductive or capacitive operating proximity sensor, as is known in the prior art. However, it is also conceivable to use an optically operating sensor for the centering operation, which, for example, directs a light beam onto the tooth to be measured and detects the light reflected by the tooth, or which detects the interruption of a light beam by the tooth to be measured during the rotation of the tooth to be measured around the workpiece axis Cl. It is furthermore conceivable that one or more further sensors are arranged on the centering probe 24, which can detect process data directly on the workpiece, for example as proposed in US 6,577,917 Bl. Such further sensors can comprise, for example, a second centering sensor for a second tooth, a temperature sensor, a further solid-borne sound sensor, a pneumatic sensor, etc.

[0086] Furthermore, in Fig. 2 an acceleration sensor 18 is indicated purely symbolically. The acceleration sensor 18 serves to characterize the vibrations of the stator of the tool spindle 15 that occur when grinding the workpiece and when dressing the grinding worm. In practice, the acceleration sensor is not arranged on a housing part, as is indicated in Fig. 2 , but is arranged, for example, directly on the stator of the drive motor of the tool spindle 15. Acceleration sensors of the type mentioned are known per se.

[0087] A coolant nozzle 19 directs a coolant jet into the machining zone. In order to record the noise transmitted via the coolant jet, a further acoustic sensor, not shown, can be provided.

[0088] In Fig. 3 , a detail III in Fig. 1 is shown enlarged. Here, the dressing device 30 can be seen particularly clearly. The dressing spindle 32 is arranged on a pivoting drive 31, which can be pivoted about an axis C4, and a disc-shaped dressing tool 33 is clamped on the dressing spindle. Instead of this or in addition, a stationary dressing tool, in particular a so-called head dresser, can also be provided, which is arranged for engagement with only the head regions of the worm thread of the grinding worm in order to dress these head regions.

[0089] Machining of a workpiece batch

[0090] For machining of workpieces which have not yet been machined, the workpieces are clamped by means of an automatic workpiece changer on workpiece spindles which are located in a workpiece changing position. The workpiece change takes place in time parallel to the machining of a further workpiece on a further workpiece spindle which is located in a machining position. When the new workpiece to be machined is clamped and the machining of the further workpiece is completed, the workpiece carrier 20 is pivoted by 180° about the C3 axis, so that the spindle with the new workpiece to be machined enters the machining position. Prior to and / or during the pivoting process, a centering operation is carried out by means of the associated centering probes. For this purpose, the workpiece spindle 21 is brought into rotation and the position of the tooth gap of the workpiece 23 is measured by means of the centering probes 24. On the basis thereof, the angle of engagement is determined. Furthermore, by means of the centering probes, an indication of an excessive variation of the tooth thickness dimension and other pre-machining errors can already be derived prior to the start of the machining.

[0091] When the workpiece spindle carrying the workpiece 23 to be machined has reached the machining position, the workpiece 23 is brought into collision-free engagement with the grinding worm 16 by moving the tool holder 12 along the X axis. The workpiece 23 is now machined in roll engagement by means of the grinding worm 16. During the machining, the workpiece is continuously fed along the Z axis with constant radial X feed. Furthermore, the tool spindle 15 is continuously displaced slowly along the displacement axis Y, in order to be able to use continuously unused regions of the grinding worm 16 during the machining (so-called displacement movement). Once the machining of the workpiece 23 is completed, the workpiece is optionally measured in-line by means of the probe 17.

[0092] In time parallel to the machining of the workpieces, the machined workpieces are removed from the further workpiece spindle and a further blank is clamped on the spindle. In each pivoting of the workpiece carrier about the C3 axis, prior to the pivoting or within the pivoting time, i.e. time-neutral, the selected components are monitored and the machining process is continued only when all defined requirements are met.

[0093] If, after machining of a certain number of workpieces, the use of the grinding worm 16 has developed in such a way that the grinding worm is too blunt and / or the tooth face geometry is too imprecise, the grinding worm is reconditioned. For this purpose, the workpiece carrier 20 is pivoted by ± 90°, so that the reconditioning device 30 enters its position opposite the grinding worm 16. The grinding worm 16 is now reconditioned by means of the reconditioning tool 33.

[0094] Data detection for process monitoring

[0095] The process monitoring device 44 serves to monitor the finish machining process carried out on the gear grinding machine 1 and, if necessary, to automatically identify and remove incorrectly machined workpieces and / or to intervene in the finish machining process in a corrective manner.

[0096] For this purpose, the process monitoring device 44 receives a large number of different measurement data from the control computer 42 on the one hand. This measurement data includes sensor data which is detected directly by the control computer 42 as well as data which the control computer 42 reads out of the spindle modules 41, for example data which indicates the current or power consumption in the tool spindle and the workpiece spindle. For this purpose, the process monitoring device can be connected to the control computer 42 via interfaces which are known per se, for example via the known Profinet standard.

[0097] The process monitoring device 44 can also have its own analog and / or digital sensor inputs in order to receive sensor data as measurement data directly from further sensors. These further sensors are generally sensors which are not required directly for controlling the actual machining process, for example acceleration sensors in order to detect vibrations or temperature sensors.

[0098] For the following discussion it is exemplarily assumed that the process monitoring device 44 detects at least the following measurement data:

[0099] • the instantaneous angular velocity or rotational speed of the tool spindle 15

[0100] • the instantaneous angular velocity or rotational speed of the workpiece spindle 21

[0101] • the current consumption or power consumption of the tool spindle 15

[0102] • the linear acceleration of the housing of the tool spindle 15 in three different spatial directions

[0103] The process monitoring device 44 can of course also detect a plurality of further measurement data.

[0104] The process monitoring device 44 detects the measurement data continuously during machining of the workpiece. In particular, the current or power consumption of the tool spindle 15 is detected at a sufficiently high sampling rate, so that there is at least one value of the power consumption during machining of each tooth surface, preferably a plurality of values per tooth surface.

[0105] Normalization operation

[0106] In the process monitoring device 44, the detected values of the current or power consumption of the tool spindle are first of all, if necessary, subjected to a filtering, for example a low-pass or band-pass filtering, in order to reduce high-frequency noise. The (possibly filtered) values are then subjected to a normalization operation. The result of the normalization operation is a normalized power indicator. The value of the normalized power indicator is calculated from the determined current or power consumption by multiplication with a normalization factor N P The normalization factor takes account of the geometrical parameters of the finishing tool, the geometrical parameters of the workpiece and the set data of the finishing machine, such as the rotational speed of the tool spindle, the radial and axial feed per revolution of the workpiece and the resulting engagement ratio between tool and workpiece.

[0107] This is based on the consideration that the current or power consumption of the workpiece spindle depends to a large extent on the geometric parameters of the finishing tool, the geometric parameters of the workpiece and the set data of the finishing machine. Thus, for example, for a ground worm with a larger diameter, under otherwise identical machining conditions, due to the longer effective lever arm, a greater torque is required and, in turn, a greater current consumption is to be expected than for a ground worm with a smaller diameter. Also, for example, under otherwise identical conditions, for a higher axial feed speed or a larger radial infeed, a correspondingly higher current consumption of the tool spindle is to be expected, the same applies for a higher rotational speed of the tool spindle. The normalization factor takes this into account. Thereby, the normalized power indicator is generally no longer or to a significantly smaller extent related to such influences than in the case of a direct detection of the current or power consumption. Since these influences have already been taken into account when calculating the normalized power indicator, deviations from the desired process can be more easily identified with the aid of the normalized power indicator than in the case of a direct measurement of the current or power consumption.

[0108] This should be done in accordance with Fig. 4 is explained in more detail. Fig. 4 Two typical variation curves 61, 62 of the power consumption of the tool spindle during the grinding of a single tooth surface are shown. The variation curve 61 is measured for a relatively large radial infeed, the variation curve 62 is measured for a significantly smaller radial infeed, under otherwise identical machining conditions. The two variation curves have a similar form: after a run-in phase, the current has an approximately constant value before the current drops again in the run-out phase. However, the current amplitudes of the current differ greatly.

[0109] In the course of the process monitoring, the time variation curve of the current consumption is continuously analyzed in order to identify impermissible process deviations. This can be done in different ways. One possibility consists, for example, in defining an envelope curve below or above which the current consumption is not permitted. If this envelope curve is undershot or exceeded, an impermissible process deviation can be inferred accordingly. Such an envelope curve 63 is exemplarily plotted in extremely simplified form in Fig. 4 . In the course of the machining with radial infeed, during which the current variation curve 62 was recorded, the envelope curve 63 represents an upper limit for the current which is not permitted to be exceeded. However, if a greater radial infeed is set, this envelope curve would no longer be necessary, as is very intuitively illustrated by the current variation curve 61: in the case of a greater radial infeed, the current consumption would exceed the envelope curve 63 even in the case of a proper machining. In correspondence therewith, the envelope curve 63 would have to be determined anew for each change in the radial infeed. This would have to be done on the basis of sample machining or empirical values. Both of these are time-consuming and prone to error.

[0110] Therefore, in the process monitoring, the measured values of the detected current or power consumption are normalized. The normalization takes into account, inter alia, the radial feed. Thereby, the normalized measured values can be directly compared with one another independently of the value of the radial feed. Accordingly, the same envelope curve can always be used in the case of different values of the radial feed. The envelope curve needs to be determined only once and can then be used for a plurality of different machining situations.

[0111] Similar considerations also apply to other methods of analyzing the measured values, for example in the case of continuous monitoring of specific spectral components of the measured values in the frequency domain.

[0112] The influence of the ground diameter of the ground worm on the detected measured values is particularly important, since the ground diameter of the ground worm changes in each dressing process. This is illustrated in Fig. 5 In Fig. 5 the maximum current consumption i_max of the tool spindle is shown for a plurality of workpieces, which were each machined by means of a gear grinding process. The workpieces were each machined by means of the same grinding worm, wherein the grinding worm was dressed after machining a specific number of workpieces. The ground diameter of the grinding worm was reduced in each dressing process. Thereby, the workpieces were machined to have a variable ground diameter. In Fig. 5 the ground diameter of the grinding worm is plotted along the horizontal axis and the maximum current consumption of the workpiece machined at this ground diameter is plotted along the vertical axis. The directly measured maximum current consumption is marked with a square. It is easily recognizable that this current consumption also decreases as the ground diameter decreases. Thereby, the measured values of the maximum current consumption at different ground diameters cannot be directly compared with one another. The normalized maximum current consumption, which is produced by applying a normalization to the measured maximum current consumption, is marked with a cross. The normalization takes into account the variable ground diameter of the grinding worm. Therefore, the value of the normalized maximum current consumption no longer depends on the ground diameter of the grinding worm.

[0113] The execution of the normalization operation is preferably carried out in real time during the machining of the workpiece in the finishing machine 1. Thereby, on the one hand, the normalized measurement values can be analyzed in real time during the machining of the workpiece, and also an impermissible process deviation can be identified immediately before or immediately after the end of the machining. The relevant workpiece can be identified and removed in real time immediately accordingly. On the other hand, it is ensured that the characteristic variable of the machining process for the respective workpiece can be calculated from the normalized measurement values directly after the end of the machining of the workpiece. In this way, the calculated characteristic variable is available immediately after the end of the machining of the workpiece. Thereby, on the one hand, further process deviations can be identified early by analyzing the calculated characteristic variable. On the other hand, there is no need to store the directly detected measurement values, i.e. the raw data, for a longer period of time as is the case in the case of offline evaluation. Rather, it is sufficient to store the calculated characteristic variable together with the signature of the respective workpiece and the selected process parameters. The storage space requirement can thereby be kept very small.

[0114] Example for calculating a characteristic variable from measured variables

[0115] In the process monitoring device 44, different characteristic variables are calculated from the (preferably normalized) power indicator and further measurement variables, which characterize the machined workpiece and its machining process. These characteristic variables are advantageously process-specific variables, such that they allow direct conclusions to be drawn about process deviations in the machining process. In particular, these characteristic variables enable the prediction of specific machining errors of the workpiece. The number of workpieces on which individual tooth part measurements are carried out can thereby be reduced, while process deviations can still be reliably identified early and, if necessary, interventions in a corrective manner can be made during the machining process.

[0116] The calculation of the characteristic variables from the measurement variables is illustrated below exemplarily for the following three characteristic variables:

[0117] (a) Cumulative pitch indicator I fP

[0118] (b) Wear indicator I Wear

[0119] (c) Profile shape indicator I ffa

[0120] All three characteristic variables are determined by spectral analysis of the time curve of the measurement variables during the machining of the workpiece, respectively.

[0121] (a) cumulative pitch indicator I fP

[0122] For determining the cumulative pitch indicator I fP , the spectral components of the (preferably normalized) power indicator are evaluated, which are in the range of the workpiece rotational speed n c ​

[0123] This is illustrated in Fig. 6 Fig. 3. The Fig. 6 shows the spectrum of the normalized current consumption (absolute value of the spectral component of the normalized current consumption as a function of the frequency "f"). This spectrum can be obtained by means of an FFT of the time curve of the normalized current consumption. The spectral component at the rotational speed n c of the workpiece spindle is indicated by an arrow. In order to quantify the spectral component, the spectral intensity at this frequency can be determined, or a narrow spectral region around this frequency can be integrated. The resulting variable is the cumulative pitch index I fP .

[0124] Generally, the greater the cumulative pitch error of the preformed tooth of the workpiece and / or the worse the concentricity of the workpiece, the greater the cumulative pitch index I fP . From the cumulative pitch index I fP it is therefore possible to infer the existing cumulative pitch error of the blank in the pre-treatment and / or the concentricity error, for example due to a faulty orientation of the clamping mechanism of the workpiece.

[0125] (b) wear indicator I Wear

[0126] In order to determine the wear index I Wear , the static component of the normalized power index is determined, i.e. the component below an upper frequency of, for example, 2 Hz. For this purpose, for example, the time curve of the power index can be low-pass filtered and integrated.

[0127] The wear index I Wear can be understood as a measure of the normalized cutting energy which is applied to the workpiece after all geometric influences and influences of the technical data used, such as radial infeed and axial feed, have been eliminated by means of a normalization operation. In short, it applies: the higher the value of the wear index I Wear , the more material is removed from the workpiece by the grinding worm at a given drive power. Thus, under otherwise constant conditions, a decrease in the wear index reflects a deterioration in the removal behavior of the tool on the workpiece. In this regard, a decrease in the value of the wear index I Wear indicates an increase in the wear of the tool.

[0128] (c) the profile shape indicator I ffa

[0129] Towards higher frequencies, the normalized current consumption is less and less informative. Therefore, in order to calculate the characteristic variables resulting from the separation of the process from the higher frequencies, other measured values are preferably used, for example the measured values of the acceleration sensor 18.

[0130] In order to determine the profile shape index I ffa, the spectral component of this measurement at the gear engagement frequency is evaluated. Here, the gear engagement frequency corresponds to the workpiece rotational speed multiplied by the number of teeth of the workpiece:

[0131] f z = n C · z

[0132] This is illustrated in Fig. 7 . The Fig. 7 shows the spectrum of the signal of the acceleration sensor 18 during machining of the workpiece. The spectral component at the tooth engagement frequency is indicated with an arrow. The spectral component can be quantified, for example, by integrating a narrow spectral range around the tooth engagement frequency.

[0133] The greater the deviation of the contour shape from the ideal contour shape according to the design, the greater the contour shape indicator I ffa in general. From the contour shape indicator, conclusions can thus be drawn about the contour shape deviation or the process deviation that causes this contour shape deviation.

[0134] Example for calculating further characteristic variables

[0135] The calculation of the characteristic variable is explained above on the basis of three examples. It goes without saying, however, that a plurality of other characteristic variables can also be determined.

[0136] Another example is the vibration indicator I vib . This characteristic variable is generated by integrating the absolute value of the measurement signal of the acceleration sensor in the frequency domain.

[0137] Identifying process deviations

[0138] By monitoring the change in the characteristic variable determined on a plurality of workpieces, an indicator for the deviation of the machining process from the idealized desired process can be determined. On this basis, the machining process can be adjusted accordingly in order to reduce the deviation. A comparison of the characteristic variable determined for a plurality of workpieces can also be used to define the limit values more precisely, such as the envelope curve mentioned above for real-time monitoring, in order to be able to determine the impermissible process deviations in real time with greater accuracy.

[0139] This is explained below on the basis of Fig. 8 to 11 .

[0140] In interpreting these diagrams, attention should be paid to the predetermined properties of the machining method chosen here.

[0141] It is noted on the one hand that each workpiece is machined in two passes, namely a roughing process and a finishing process. The displacement strategy is as follows: First, each workpiece is roughed with the aid of a specific grinding worm region. Then the grinding worm is displaced by a specific amount (in the diagram to higher Y values, i.e. to the left), so that a not yet used grinding worm region is used for finishing. After finishing, the grinding worm is displaced again back to the end of the grinding worm region that was used for roughing last time, and the next grinding worm region is used for roughing of the next workpiece. Thereby, almost each grinding worm region is first used for finishing a workpiece and later for roughing a later workpiece. Only the rightmost grinding worm region near Y=0 is used in this displacement strategy only for the roughing process. In Fig. 8 to 11 the machining position Y shows: Which worm region along the grinding worm width is used for machining the respective workpiece during roughing.

[0142] It is noted on the other hand that each time the grinding worm is re-dressed when reaching the end of the grinding worm upon displacement. Upon re-dressing, the outer diameter of the grinding worm is reduced accordingly. Thereby, on the one hand the lever ratio changes, which converts the drive torque into the cutting force upon grinding a workpiece, and on the other hand, the engagement ratio changes during machining each tooth surface as well. In Fig. 8 to 11 the characteristic variables are shown over multiple dressing cycles as a function of the machining position Y at roughing, respectively. Due to the normalization operation, the dressing has no or little influence on the respective characteristic variables. The reason for this is that the influence of the dressing process on the geometry of the grinding worm is taken into account by the normalization factor. Thereby, the values of the normalized power index over different dressing cycles can be directly compared with each other. Accordingly, the characteristic variables shown for different dressing cycles can be directly compared with each other, although the outer diameter of the grinding worm is variable. There is a big advantage of the proposed normalization operation in this respect.

[0143] In Fig. 8 the cumulative pitch index l fP is shown for the roughing process of multiple workpieces over multiple dressing cycles as a function of the machining position Y along the grinding worm width. In this diagram, it is distinguished whether a workpiece is machined on the first workpiece spindle or on the second workpiece spindle (l fP (C1) or l fP (C2)), respectively. The cumulative pitch index l fP (C1) of workpieces machined with the aid of the first workpiece spindle is shown with triangles, while the cumulative pitch index l fP (C2) of workpieces machined with the aid of the second workpiece spindle is shown with crosses.

[0144] It can be immediately recognized that the cumulative pitch indicator for the first workpiece spindle is significantly higher on average than the cumulative pitch indicator for the second workpiece spindle. This indicates a concentricity error of the workpiece on the first workpiece spindle due to a false orientation of the clamping mechanism in the case of identical pre-machining of the workpiece. This concentricity error can cause an undesirable noise generation when using the toothings thus produced. At the same time it can be recognized that the finishing process has virtually no influence on the value of the determined cumulative pitch indicator as a result of the normalization operation.

[0145] The operator is made easy to recognize the concentricity error and the corresponding cause by the way that the characteristic variable I fP is associated with another variable of the machining process, in this case with the position Y along the grinding worm width and is shown in a visual manner. Instead of being associated with the position Y, it can also be associated here with other variables, in the simplest case with the consecutive workpiece number.

[0146] In Fig. 9 the profile shape indicator 1 ffa is shown for a plurality of workpieces again as a function of the machining position Y of the roughing process over a plurality of finishing cycles. It can be recognized that, independently of the finishing cycle, the profile shape indicator is significantly smaller at the right edge of the grinding worm (close to Y = 0), i.e. at the position where the machining is started after each finishing process, and increases under a large scattering width towards the left edge of the grinding worm (Y = 40 mm). The diagram thus shows that the respective first workpiece of the finishing cycle is produced with the correct profile shape, while for the subsequent workpieces an increasingly greater profile shape deviation occurs, which also scatters to a large extent. If the shifting strategy shown above is elucidated, it shows that the grinding worm is subjected to excessive loads in the region in which the finishing process is first carried out by means of the grinding worm and the roughing process is carried out later.

[0147] In Fig. 9 the profile shape error and the corresponding cause are then made easy for the operator to recognize by the way that the characteristic variable I ffa is associated with another variable of the machining process, in this case again with the position Y along the grinding worm width and is shown in a visual manner.

[0148] In Fig. 10 the wear indicator 1 WearAgain as a function of the machining position Y of the roughing process. The wear indicator has a relatively large value at the right end of the grinding worm near Y = 0. In further extension of the grinding worm width, the wear indicator quickly drops to significantly lower values. It is noted here that the wear indicator is not a direct measure of the wear itself, but rather a measure of the amount of material removed on the tooth surface. Thus, on the largest areas of the grinding worm width, less material is removed from the tooth surface than in the area near the right side at Y = 0. This indicates that the wear increases in all areas except the area of the right side. The extension of the wear indicator over the grinding worm width confirms in this regard the insight that the wear indicator shows that the grinding worm is excessively worn at any location except the area of the right side near Y = 0, which can also be obtained from the extension of the profile shape indicator over the grinding worm width.

[0149] In Fig. 11 , the vibration indicator is I vib . Again for a plurality of workpieces as a function of the machining position Y of the roughing process. The vibration indicator shows that a higher vibration load is generated in the worn area of the grinding worm, although the material removal rate is lower. This in turn can cause undesirable noise when using a tooth part manufactured in this way. The extension of the vibration indicator over the grinding worm width confirms in this regard the insight gained from the extensions of the profile shape indicator and the wear indicator.

[0150] In order to intervene in a corrective manner here, for example the rotational speed of the tool spindle, the radial feed per revolution of the workpiece or the axial feed can be reduced.

[0151] Comparison with measured values from tooth portion measurements

[0152] The determined characteristic variable can be compared to the results of tooth part measurements on a tooth part test bench for the selected workpieces. In this way, parameters describing the correlation of the characteristic variable to the actually existing shape deviations can be determined quantitatively. For example, in the case of a linear correlation between the characteristic variable and the shape deviation, a linear regression can be carried out in order to determine the parameters of the linear correlation. It is thereby possible to quantify the shape deviations directly for each machined workpiece by means of the characteristic variable, which would otherwise only be possible by means of tooth part measurements and would represent a disproportionate effort.

[0153] Web-based interface

[0154] The graphical display of the determined characteristic variable and its correlation to other characteristic variables of the machining process can in particular be made platform-independent on any client computer via a web browser. Other evaluation measures can also be implemented accordingly platform-independently. Remote analysis is thereby also facilitated.

[0155] Automatic identification of machining errors

[0156] In the above example, the spread of the different characteristic variables over the grinding worm width is visually analyzed by the operator on the machine or by an expert on an arbitrary client computer. However, instead of this, such an analysis can also be performed fully automatically.

[0157] For this purpose, the process monitoring device 44 can execute an algorithm which automatically identifies patterns in the characteristic variables determined on the plurality of workpieces. For this purpose, in particular machine learning algorithms are suitable, as are known in various embodiments. Such algorithms are often also referred to as "artificial intelligence". An example thereof is a neural network algorithm. It is clear that such an algorithm in the above example can easily identify, for example, the difference in the cumulative pitch index between the first and second workpiece spindle or the above-described wear behavior over the grinding worm width. For this purpose, the algorithm can generally be trained by means of a training data set. The training data set can take into account, in particular, parameters which describe the correlation of the characteristic variables with the actually existing shape deviations from the tooth measurement.

[0158] Now the necessary measures can be taken to eliminate the process deviations. For example, upon identification of a concentricity error, the centering of the workpiece clamping mechanism on the respective workpiece spindle can be corrected manually or automatically. In the case of an excessive wear, the radial infeed and / or the axial feed can be reduced accordingly. These measures can also be carried out manually or automatically.

[0159] Force model

[0160] The normalization factor is preferably calculated in a model-based manner.

[0161] In the literature, there are models for gear grinding which describe the correlation of the cutting forces with the geometric parameters of the tool and the workpiece and with the technical parameters. See, by way of example, the above-mentioned work by H. Schriefer et al., "Continuous Generating Gear Grinding", Reishauer AG self-published, Wallisellen 2010, ISBN 978-3-033-02535-6, chapter 4.7.3 "Cutting Force", pages 319-322.

[0162] In the following, reference is made to a force model which is described in the thesis "Numerische Simulation deskontinuierlichen "Unter Berücksichtigung des dynamischen Verhaltens des Systems Maschine-Werkzeug-Werkstück", dissertation ETH Zurich, No. 24172, https: / / doi.ora / 10.3929 / ethz-b-000171605 A method for numerically modeling a continuous gear grinding process is disclosed in the above-mentioned dissertation. To this end, a model for calculating the cutting forces is proposed, and a method for experimentally determining the parameters of the model by means of measurements is proposed.

[0163] The force model for the normal force is given in equation 4.27 in the dissertation of C. Dietz as follows:

[0164]

[0165] Here, α e denotes the depth of cut, α p denotes the width of cut, l k denotes the contact length of the grinding zone, v c denotes the cutting speed, and v f denotes the feed speed. The constants F0and k as well as the exponents ε1and ε2are variables to be determined empirically.

[0166] The cutting force F c is proportional to the normal force F n , wherein the proportionality constant μ is called the force ratio:

[0167] F c = μ F n

[0168] The force ratio μ is again a variable to be determined empirically.

[0169] The geometric variables α e , α p and l k describe the grinding zone. The geometric variables can be calculated analytically, as is explained in chapter 4.5.1 of the dissertation of C. Dietz, or determined numerically by means of a penetration calculation.

[0170] In particular, for the contact length l k the following analytical relationship can be derived:

[0171]

[0172] Here, r pSS,eq denotes the equivalent grinding worm radius, q t denotes the nominal allowance, and s zindicates the radial feed per revolution of the workpiece. The equivalent grinding worm radius r pSS,eq from the actual grinding worm radius r pSS is obtained as follows:

[0173]

[0174] Here, a ss indicates the engagement angle of the grinding worm toothing.

[0175] For the cutting width a p the following analytical relationship can be derived:

[0176]

[0177] Here, a p,max indicates the maximum cutting width. The maximum cutting width is obtained as follows:

[0178]

[0179] Here, L y indicates the roll length.

[0180] The cutting depth corresponds to the nominal allowance, which in turn is related to the radial feed Δx as follows:

[0181] a e = q t = Δx • sin(a t )

[0182] Here, a t indicates the engagement angle of the workpiece.

[0183] The cutting speed v c and the feed speed v f result from the kinematics of the gear grinding process. As explained in Chapter 4.7.1 of the dissertation by C. Dietz, analytical formulas can also be given for these variables. Thus for the cutting speed v c the following applies:

[0184]

[0185] Here, v cu indicates the peripheral speed, v cα the axial speed component, and v cw the roll speed.

[0186] For the peripheral speed v cu the following applies:

[0187] v cu = n B πd pss / 60

[0188] Here, n B denotes the rotational speed of the tool in U / min, while d pSS denotes the diameter of the ground worm at the instantaneous point of contact along the profile height of the respective worm thread.

[0189] For the axial velocity component v ca apply:

[0190]

[0191] Here, m nSS denotes the normal module of the ground worm, z0denotes the number of threads, and γ denotes the pitch angle of the worm tooth.

[0192] For the roll speed v cw apply:

[0193]

[0194] Here, α SSy denotes the local angle of engagement of the point of contact on the ground worm.

[0195] For the feed speed v f apply:

[0196]

[0197] Here, β denotes the helix angle of the tooth on the workpiece.

[0198] Alternatively, the geometric variables α e , α p and l k as well as the cutting speed v c and the feed speed v f for describing the cutting zone can also be determined from numerical process simulations.

[0199] The constants F0and k, the exponents ε1and ε2as well as the friction ratio μ can be determined empirically, as this is exemplarily explained in Chapter 5.3 of the thesis by C. Dietz.

[0200] In the thesis by C. Dietz, the following values were determined empirically for the gear grinding of a tooth made of hardened steel with a tool with ceramic bond using corundum as abrasive:

[0201] F0= 21.2422 N

[0202] k = 4.4338

[0203] ε1= 0.1950

[0204] ε2 = 2.0136

[0205] μ = 0.3

[0206] With these values the actually measured process forces can be reproduced with very high accuracy by the force model.

[0207] For other material pairs the values of the mentioned variables can differ from the values given above. However, by comparing the measured and calculated force values these values can easily be determined empirically.

[0208] Modeling of process power

[0209] Power to be applied by the tool spindle P B From the cutting force F c and the product of the circumferential speed v of the grinding worm at the contact point between the grinding worm and the workpiece cu On the one hand, the circumferential speed is proportional to the rotational speed n of the tool spindle B and on the other hand to the effective lever arm, which corresponds to half the diameter d of the grinding worm at the contact point. pSS

[0210] The process power can be modeled overall as:

[0211] P- B = F c • n B π d pSS / 60

[0212] Here, the diameter d of the grinding worm at the contact point pSS can be replaced in a good approximation by its outer diameter d aSS .

[0213] Calculation of normalization factors

[0214] Based on this process power model, a normalization factor can for example be chosen as follows:

[0215]

[0216] Here, d αSS,max denotes the maximum outer diameter present after the first dressing process. Alternatively, any other reference value can also be used here. The expression n B,ref denotes an arbitrary reference rotational speed.

[0217] In this model, for the exponents E1, E2 and E3 applies:

[0218] E1 = E2 = E3 = 1

[0219] ​However, in an extended version of the model, these exponents can also deviate from 1 and can be determined empirically.

[0220] The first factor (with the exponent E1) is generated directly from the force model. It takes into account the geometry and the technical specifications of the grinding worm and the workpiece, in particular the radial infeed and the axial feed.

[0221] The second factor (with the exponent E2) takes into account the lever ratio at the contact point, which varies as a function of the grinding worm diameter.

[0222] The third factor (with the exponent E3) takes into account the correlation of the process power with the rotational speed of the tool spindle.

[0223] As results from the above discussion of the force model and from further embodiments in the C. Dietz paper, the cutting force changes to some extent during the tooth surface machining. However, for the purpose of process monitoring, the cutting force can be considered constant during the tooth surface machining, neglecting run-in and run-out. Accordingly, the normalization factor N P is recalculated after each modification of the grinding worm geometry and of the technical parameters, in particular the radial infeed and / or the axial feed. P It can then be used for all workpieces of the grinding cycle.

[0224] It is clear that the normalization factor N P can also be calculated in a different way than proposed above. More complex normalization operations are also conceivable, for example operations which first comprise a subtraction in order to eliminate an offset and then a multiplication or division.

[0225] The above considerations apply to gear grinding. For other methods of finish machining, there are different cutting force models, and accordingly, for other methods of finish machining, the normalization factor is different from the normalization factor explained above.

[0226] Flowchart of exemplary method

[0227] Fig. 12 A flow chart showing an exemplary method for process monitoring in the gear grinding machining of a batch of similar workpieces by means of a finish machining machine of the type shown in Fig. 1 A flow chart showing an exemplary method for process monitoring in the gear grinding machining of a batch of similar workpieces by means of a finish machining machine of the type shown in

[0228] In step 110 the finishing machine is set up and the relevant process parameters, in particular the geometrical and technical parameters of the grinding worm and the workpiece, such as the radial infeed and the axial feed, are entered into the machine control 42 via the operating panel 43. In step 111 the dressing grinding worm 16 is dressed and the outer diameter of the dressed grinding worm is determined. In step 112 a normalization factor is calculated on the basis of the process parameters and the now existing grinding worm outer diameter.

[0229] In block 120 the individual workpieces of the batch are machined. During the machining, in step 121 the measured variables are continuously detected by the process monitoring device 44. In step 122 at least some of the measured variables, in particular those relating to the power consumption of the tool spindle, are normalized in real time. The now partially normalized measured variables are continuously analyzed in real time in step 123 in order to identify possible machining errors directly online from the process deviations, also during the machining of the individual workpieces. If a possible machining error is identified, a corresponding indication variable is set in the process monitoring device. Steps 121 to 123 are repeated continuously during the machining of the workpieces.

[0230] Immediately after the end of the machining of the workpieces, in step 124 characteristic variables are calculated from the partially normalized measured variables. The characteristic variables are compared to the specifications. If it is derived that the characteristic variables deviate too much from the specifications, the indication variable for machining errors is set again.

[0231] In step 125 the workpiece handling system is indicated by means of the indication variable to remove the workpieces identified as indicating machining errors from the further process. These workpieces can be subjected to a more detailed examination or immediately discarded as NIO parts.

[0232] In step 126 a data set is stored in a database for each workpiece. The data set comprises the unambiguous workpiece identification, the most important process parameters, the determined characteristic variables and possibly the indication variable.

[0233] The machining of the workpieces is now repeated in the same way until the grinding worm is worn to the extent that a new dressing operation is required. In this case step 111 is repeated, i.e. the grinding worm is re-dressed and its new outer diameter is determined. Correspondingly, the normalization factor is recalculated in step 112. Now the machining 120 of the workpieces is continued with the re-dressed grinding worm and the new normalization factor.

[0234] Fig. 13Figuratively shown: How further process deviations can be automatically identified from stored data sets for a plurality of workpieces. In step 131, data sets for a plurality of workpieces are read from a database. In step 132, the data sets are analyzed by means of a KI algorithm (KI = artificial intelligence) in order to recognize process deviations from the data sets, which can not be directly recognized during machining of the individual workpieces. In step 133, based on the results of the analysis, measures for correcting the machining process are automatically initiated (for example, reduction of axial feed, etc.). These steps can be performed as soon as a certain minimum number of workpieces has been machined. However, the analysis can also be performed subsequently after the machining of the workpiece batch has ended in order to, for example, subsequently recognize those workpieces that were affected by machining errors.

[0235] The process requires only moderate computing and storage requirements, since the stored data sets are very small compared to the amount of data directly detected during machining.

[0236] As is figuratively shown in Fig. 14 Irrespective thereof, the data sets can be read at any point in time from the database by means of a client computer via a network (step 141) and processed and output in a graphic manner (step 142). This process also requires only very moderate computing and storage requirements. This makes it possible for the process to be performed by means of a plug-in in a web browser. Based on the output, an operator can perform an error analysis and, for example, re-determine the above-mentioned envelope curve applied during real-time analysis.

[0237] Block diagram of functional blocks for a process monitoring device

[0238] In Fig. 15 A block diagram is shown exemplary in

[0239] A normalization calculation device 410 calculates normalization factors, if necessary. A detection device 420 serves for detecting measurement values. A normalization device 430 normalizes at least a portion of the measurement values immediately after they have been detected. An error recognition device 440 analyzes the portion of the measurement values that have been normalized and recognizes unallowable process deviations therefrom. Based thereon, a processing device 441 (which does not belong to the process monitoring device as such) removes workpieces that have been machined with unallowable process deviations. After the machining of the workpieces has ended, a characteristic variable calculation device 450 calculates characteristic variables from the portion of the measurement values that have been normalized. A data communication device 460 serves for communication with a database server. A deviation determination device 470 serves for automatic determination of process deviations. To this end, the deviation determination device has a processor device 471 which executes a KI algorithm.

[0240] It goes without saying that numerous modifications to the above examples are possible.

Claims

1. A method for monitoring a machining process, wherein the tooth surface of a pre-formed tooth workpiece (23) is machined by means of a finishing machine (1), wherein the finishing machine (1) has a tool spindle (15) for rotatably driving a finishing tool (16) about a tool axis (B) and at least one workpiece spindle (21) for rotatably driving the pre-formed tooth workpiece (23), characterized in that, The method includes: During the machining engagement between the finishing tool (16) and the workpiece (23), multiple measurements are detected, including the value of a power index that shows the instantaneous power consumption of the tool spindle (15). Apply a normalization operation to the value of the power index or the variable derived therefrom to obtain a normalized value. The normalization operation is associated with at least one geometric parameter, wherein the at least one geometric parameter is selected from the geometric parameters of the finishing tool (16) and the geometric parameters of the workpiece (23). Analyze the normalized values ​​to determine unacceptable process deviations in the detection; Change at least one of the geometric parameters; and The normalization operation for the geometric parameters is recalculated after the change.

2. The method according to claim 1, wherein the normalization operation is performed in real time, and the finishing tool (16) is engaged with the workpiece (23).

3. The method of claim 2, wherein the normalized value is analyzed in real time to determine unacceptable process deviations.

4. The method according to claim 3, comprising: Remove the following workpieces for which unacceptable process deviations have been determined.

5. The method according to claim 1, comprising: The characteristic variables of the machining process are calculated from the measured values ​​or values ​​derived therefrom, wherein at least one of the characteristic variables is associated with a predefined machining error of the workpiece (23).

6. The method according to claim 5, wherein the method comprises: Perform tooth measurements on the selected workpiece (23) to determine at least one tooth measurement variable characterizing the predefined machining error for each workpiece; and Determine a correlation parameter, which characterizes the correlation between the at least one feature variable and the at least one tooth measurement variable.

7. The method of claim 5, wherein the calculation of at least one of the characteristic variables comprises spectral analysis of the measured value or a value derived therefrom to obtain a plurality of spectral components.

8. The method of claim 7, wherein the spectral components are evaluated at multiples of the rotational speed of the tool spindle (15) and / or the workpiece spindle (21).

9. The method according to claim 7, wherein the processing is a rolling process, wherein the finishing tool (16) and the workpiece (23) are engaged by rolling, and wherein the characteristic variable includes at least one of the following variables: The cumulative pitch index is calculated from the spectral components of the spindle (15) at its rotational speed and is associated with the cumulative pitch error or concentricity error of the workpiece (23). Wear index, wherein the wear index is calculated from low-frequency spectral components and is related to the degree of wear of the finishing tool (16); and The profile shape index is calculated from the spectral components at the gear engagement frequency and is associated with the profile shape deviation of the workpiece (23).

10. The method according to claim 5, wherein the method comprises: The dataset is stored in a database (46), wherein the dataset includes explicit markings of the workpiece, at least one geometric parameter, and at least one of the feature variables.

11. The method according to claim 5, wherein the method comprises: An analysis is performed on the values ​​of at least one of the characteristic variables used for multiple workpieces (23) in order to determine process deviations; and The processing procedure is modified to reduce the process deviation.

12. The method of claim 11, wherein the analysis is performed using a trained machine learning algorithm.

13. The method of claim 11, wherein the analysis comprises: The value of at least one of the characteristic variables used for multiple workpieces (23) is associated with other variables of the machining process.

14. The method according to claim 5, wherein the method comprises: Graphically output the value of at least one of the feature variables used for multiple workpieces, or the value derived from it.

15. The method according to claim 1, The recalculation of the normalization operation involves applying a model that describes the expected correlation between the measured values ​​and the geometric parameters.

16. The method according to claim 15, wherein, The model describing the expected correlation between the measured value and the geometric parameter is a model of process force or process power.

17. The method of claim 15, wherein the recalculation of the normalization operation includes compensation for changes in dimensions of the finishing tool (16).

18. The method according to claim 17, wherein, The changed dimension of the finishing tool (16) is its outer diameter.

19. A method for monitoring a machining process, wherein the tooth surface of a pre-formed tooth workpiece (23) is machined by means of a finishing machine (1), wherein the finishing machine (1) has a tool spindle (15) for rotatably driving a finishing tool (16) about a tool axis and at least one workpiece spindle (21) for rotatably driving the pre-formed tooth workpiece (23), wherein the machining process is a continuous roll grinding process, wherein the finishing tool (16) is a grinding worm and the finishing tool (16) and the workpiece (23) are roll engaged, characterized in that, The method includes: During the machining engagement between the finishing tool (16) and the workpiece (23), multiple measurements are detected, including a power index that displays the instantaneous power consumption of the tool spindle (15); and Calculate at least one characteristic variable of the processing from the detected measurements. The at least one feature variable is associated with a predefined machining error of the workpiece (23).

20. The method of claim 19, wherein the method comprises: Perform tooth measurements on the selected workpiece (23) to determine at least one tooth measurement variable characterizing the predefined machining error for each workpiece; and Determine a correlation parameter, which characterizes the correlation between the at least one feature variable and the at least one tooth measurement variable.

21. The method of claim 19 or 20, wherein the calculation of at least one of the characteristic variables comprises spectral analysis of the measured value or a value derived therefrom to obtain a plurality of spectral components.

22. The method of claim 21, wherein the spectral components are evaluated at multiples of the rotational speed of the tool spindle (15) and / or the workpiece spindle (21).

23. The method of claim 21, wherein the characteristic variable includes at least one of the following variables: The cumulative pitch index, wherein the cumulative pitch index is calculated from the spectral components at the rotational speed of the workpiece spindle (15) and is associated with the cumulative pitch error or concentricity error of the workpiece (23); and The profile shape index is calculated from the spectral components at the gear engagement frequency and is associated with the profile shape deviation of the workpiece (23).

24. The method according to claim 19 or 20, wherein the method comprises: The dataset is stored in a database (46), wherein the dataset includes a specific identifier of the workpiece, at least one process parameter, and at least one of the feature variables.

25. The method according to claim 19 or 20, wherein the method comprises: An analysis is performed on the values ​​of at least one of the characteristic variables used for multiple workpieces (23) in order to determine process deviations; and The processing procedure is modified to reduce the process deviation.

26. The method of claim 25, wherein the analysis is performed using a trained machine learning algorithm.

27. The method of claim 25, wherein the analysis comprises: The value of at least one of the characteristic variables used for multiple workpieces (23) is associated with other variables of the machining process.

28. The method according to claim 19 or 20, wherein the method comprises: Graphically output the value of at least one of the feature variables used for multiple workpieces, or the value derived from it.

29. A finishing machine for machining the tooth surface of a workpiece with pre-formed teeth, the finishing machine comprising: Tool spindle (15) for rotating and driving the finishing tool (16) around the tool axis; At least one workpiece spindle (21) for rotating a workpiece (23) with pre-made teeth; A control device (40) for controlling the machining process of the workpiece (23) using the finishing tool (16); and A process monitoring device (44) is configured to perform a method for monitoring the processing process according to any one of claims 1 to 28.

30. The finishing machine according to claim 29, wherein the process monitoring device (44) comprises: Detection device (420) for detecting multiple measurements during the machining engagement of the finishing tool (16) and the workpiece (23); and A normalization device (430) is used to apply a normalization operation to at least a portion of the measured values ​​or the values ​​of variables derived from the measured values ​​in order to obtain normalized values. The normalization operation is associated with at least one process parameter, wherein the at least one process parameter is selected from the geometric parameters of the finishing tool (16), the geometric parameters of the workpiece (23), and the setting parameters of the finishing machine (1).

31. The finishing machine according to claim 29, wherein the detected measurement includes a power index that shows the instantaneous power consumption of the tool spindle (15), and wherein the normalization operation is applied to the value of the power index or a variable derived therefrom.

32. The finishing machine according to claim 30 or 31, wherein the normalization device (430) is configured to perform the normalization operation in real time during the machining engagement of the finishing tool (16) and the workpiece (23).

33. The finishing machine according to claim 32, wherein the process monitoring device has an error identification device (440) configured to analyze the normalized value in real time in order to determine unacceptable process deviations.

34. The finishing machine according to claim 33, the finishing machine comprising a workpiece handling device (441) configured to automatically remove workpieces for which unacceptable process deviations have been determined.

35. The finishing machine according to claim 30 or 31, wherein the process monitoring device has a feature variable calculation device (450) for calculating feature variables of the machining process from the measured values ​​or values ​​derived therefrom.

36. The finishing machine according to claim 35, wherein the feature variable calculation device (450) is configured to perform spectral analysis on the measured value or the value derived therefrom to obtain a plurality of spectral components.

37. The finishing machine according to claim 36, wherein the feature variable calculation device (450) is configured to evaluate the spectral components at multiples of the rotational speeds of the tool spindle (15) and / or the workpiece spindle (21).

38. The finishing machine according to claim 36, wherein the finishing mechanism performs a rolling process, wherein the finishing tool (16) and the workpiece (23) engage in rolling, and wherein the characteristic variable includes at least one of the following variables: The cumulative pitch index is calculated from the spectral components of the spindle (15) at its rotational speed and is associated with the cumulative pitch error or concentricity error of the workpiece (23). Wear index, wherein the wear index is calculated from low-frequency spectral components and is related to the degree of wear of the finishing tool (16); and The profile shape index is calculated from the spectral components at the gear engagement frequency and is associated with the profile shape deviation of the workpiece (23).

39. The finishing machine according to claim 35, wherein the process monitoring device comprises: Data communication device (460) for transmitting dataset to database (46), wherein the dataset includes a definite identifier of the workpiece, at least one process parameter and at least one of the feature variables.

40. The finishing machine according to claim 30 or 31, wherein the process monitoring device comprises: Deviation determination device (470) is used to determine the process deviation between the processing procedure and the desired process based on the value of at least one characteristic variable among a plurality of workpieces (23).

41. The finishing machine of claim 40, wherein the deviation determination device (460) includes a processor device (461) programmed to execute a trained machine learning algorithm to determine the process deviation.

42. The finishing machine according to claim 30 or 31, wherein the process monitoring device (44) comprises: A normalization calculation device (410) is used to recalculate the normalization operation when at least one process parameter changes.

43. The finishing machine according to claim 42, wherein the normalization calculation device (410) is configured to apply a model describing the expected correlation between the measured value and the process parameter when recalculating the normalization operation.

44. The finishing machine according to claim 43, wherein the model describing the expected correlation between the measured value and the process parameter is a model of process force or process power.

45. The finishing machine according to claim 42, wherein the normalization calculation device (410) is configured to perform compensation with respect to the dimensions of the finishing tool (1).

46. ​​The finishing machine according to claim 45, wherein the dimension of the finishing tool (1) is its outer diameter.

47. A computer-readable medium having a computer program stored thereon, the computer program comprising instructions that cause a process monitoring device in a finishing machine (1) according to any one of claims 29 to 46 to perform the method according to any one of claims 1 to 28.

Images