Method and arrangement for determining deviations in a driven motion with multiple motion cycles

Abstract

The invention relates to a method and an arrangement for determining deviations during a movement, which is carried out in particular during machining or forming of the workpiece.The method comprises: - Acquiring at least one time-resolved or position-resolved measured quantity that is variable during a movement, wherein the movement is performed by a driven object (16) and comprises several successive movement cycles; - Determining time-resolved or position-resolved setpoints of the measured quantity, wherein, based on a plurality of values ​​of the measured quantity, each acquired for the same time or movement position within the movement cycles, at least one setpoint for the respective time or movement position is determined; - Determining, for at least one time or movement position, a deviation of at least one further acquired value of the measured quantity from at least one setpoint for that time or movement position.

Description

The invention relates to a method and an arrangement for determining deviations during a movement, and in particular during a rotational process. Furthermore, the invention relates to a method and a system with an arrangement for determining deviations during machining or forming of a workpiece, e.g., milling or pressing. Movements, and especially rotational processes, are performed in various contexts, particularly in machine-driven forms. Examples include rotary transfer machines, presses, and pumps. Another example is machining, where the machining tool and / or the workpiece are driven rotaryally or axially. In machining processes, tool breakage or the breakage of individual cutting edges leads to poor surface quality and downtime. This results in additional costs, and the manufactured components must be reworked or discarded. Often, quality control of the components only takes place at the end of the production line, after the tool breakage has occurred, meaning that several other components can be damaged in the meantime. Therefore, approaches exist for the automated monitoring of machining processes and for the detection of tool and cutting edge breakage in order to save costs and production time. Damage to the forming tool can also lead to similar problems during press machining or other forming processes. One option is to inspect the tool after each machining cycle and / or machining program. This inspection can be performed in a separate measuring device, for example, optically using a laser (see, for example, BLUM LC50-DIGILOG or Renishaw TRS2) or mechanically using a probe (e.g., BLUM ZX-Speed) or sensor needle. Disadvantages of this type of inspection include the need for a separate measuring system, the fact that tool breakage is only detected after the machining or programming cycle is complete, and the requirement for a separate measurement cycle after each machining or programming cycle, which negatively impacts cycle times. Furthermore, systems exist that directly detect tool breakage events based on measurement data acquired during the process. An overview of previously published approaches can be found in the following review articles: Zhou & Xue: Review of tool condition monitoring methods in milling processes, Int. J. Adv. Manuf. Technol. 96 (2018), 2509-2523; and T. Mohanraj: Tool condition monitoring techniques in milling process - a review, J. Mater. Res. Technol. 9 (2020), 1032-1042. In such cases, the use of special additional sensors is often necessary, such as acceleration or structure-borne sound sensors. This increases maintenance effort and investment costs. When analyzing machine-internal data that utilizes existing sensors, existing algorithms are generally designed to detect tool or cutting edge breakage based on predefined limit values ​​in a current signal of the main spindle or feed axis, or derived characteristics thereof; see, for example: Li X.: Detection of Tool Flute Breakage in End Milling Using Feed-Motor Current Signatures, IEEE / ASME Trans. Mechatronics. (2001) 6:491-243. A disadvantage in these cases is that the limit values ​​must be set separately for each program cycle. Furthermore, the cutting forces and thus the motor current amplitudes for the same program cycle are subject to considerable fluctuations, for example, due to temperature dependencies, batch-dependent changes in the properties of the blanks or the cutting edges of the tools. Consequently, false alarms occur, or the limit values ​​must be set very conservatively, resulting in inaccurate fault detection. Similar problems exist with approaches in which the limit values ​​are determined based on cutting force models, see e.g. Altintas Y., Aslan D.: Integration of virtual and on-line machining process control and monitoring, CIRP Ann. (2017) 66:349-352 . Commercially available systems (e.g., MCU Toolinspect, Marposs / ARTIS CTM, Ceratizit / Komet Toolscope) enable the monitoring of machining processes primarily based on previously recorded target curves (also known as stored learning curves), see also DE 10 2016 100 503 B3. Using a previously stored reference curve of the spindle current for a specific program cycle, tolerance limits are determined, and an alarm is triggered or a program stop is forced if these limits are exceeded or fallen below. A disadvantage of this approach is that the tolerance limits must be learned for each program cycle and, due to the aforementioned batch-related variations in cutting forces, must be set conservatively. A second approach, which can be partially implemented with the aforementioned commercial systems, is the detection of an abrupt drop in spindle current to the idle level (the level of spindle current for a rotating spindle without the cutting edges engaged). The disadvantage here is that only a complete tool breakage can be detected, while breaks in individual cutting edges cannot be detected. The invention therefore aims to improve the monitoring of driven movements and, in particular, to improve the automatic and / or computer-aided monitoring of machine-executed and / or machine-driven movements, such as rotational processes. This particularly concerns improvements in the accuracy of error detection as well as in the costs and setup effort of the monitoring. This problem is solved by the attached independent claims. Advantageous further developments are given in the dependent claims, in this description, and in the figures. Accordingly, a method for determining deviations during a movement, which is carried out in particular during machining or forming of a workpiece, is defined as: - Acquiring at least one time-resolved or position-resolved measured quantity that is variable during the movement, wherein the movement is performed by a driven object (16) and comprises several successive movement cycles; - Determining time-resolved or position-resolved target values ​​of the measured quantity, wherein, based on a plurality of values ​​of the measured quantity, each of which is acquired for the same time or the same movement position within the movement cycles, at least one target value for the respective time or movement position is determined;- Determine, for at least one point in time or at least one movement position, a deviation of at least one further recorded value of the measured quantity from at least one target value for that point in time or movement position. The method thus primarily involves determining target values ​​for a measured quantity, either time-resolved or position-resolved. This is achieved by utilizing measured values ​​that were acquired at the same points in time or at the same position within repeated motion cycles. A plurality of such measured values, available for the same point in time or position within the motion cycles, serves as the basis for calculating at least one target value for that specific point in time or position. The goal is to determine target values ​​that apply either to specific points in time during each motion cycle or to specific positions of the movement, such as along a linear path or a rotation angle. In this way, for example, the measured quantity of a rotating system can be acquired for each angle within a revolution—e.g., at 90°, 180°, or 270°.The measured values ​​from several cycles are combined and the target value for a specific position or time is determined by, for example, averaging or other methods. According to further training, a given point in time within a movement cycle is defined relative to a reference point of the movement cycle, e.g., a beginning, end, or midpoint of the movement cycle, and / or to a period of the movement cycle. According to a further development, the motion process is a rotational process and the motion position is a rotational angle; or, where the motion process is axial and, in particular, linear, and where the motion position is a position along the axis of motion. An example of a linear motion is the ram travel of a press. According to a specific example, a method for determining deviations during a rotational process is proposed as an example of a movement, in particular for monitoring at least one measured quantity in connection with the rotational process, wherein the rotational process is carried out in particular during machining of a workpiece, comprising the method: - Acquiring at least one rotationally angle-resolved measured quantity that is variable during a rotational process, wherein the rotational process is carried out by a rotaryally driven object and comprises several successive revolutions as an example of movement cycles; - Determining rotationally angle-resolved target values ​​of the measured quantity, wherein, based on a plurality of values ​​of the measured quantity acquired for a respective rotational angle, at least one target value for the respective rotational angle is determined as an example of a movement position;- Determine, for at least one rotation angle, a deviation of at least one other recorded value of the measured quantity from at least one target value for this rotation angle. The following statements also apply in general terms to the first-mentioned embodiment, i.e., also to general movements and motion cycles as well as time-resolved measured values ​​and / or linear movements. The determination of deviations, and in particular any procedural step disclosed herein, can be carried out automatically and / or computer-aided. For example, these measures can be performed, at least partially, by a data acquisition and / or analysis device. These devices can each include at least one processor and / or be a machine computing device. The measurement can be acquired, for example, using sensors and / or other suitable acquisition devices. The measured values ​​can be transmitted, particularly automatically, to a processing unit for further analysis. The measurement can be variable, at least during faulty rotation processes, and / or at least vary differently than during fault-free rotation processes. These changes can, for example, occur periodically and / or repeat with each rotation. The rotary-driven object is not restricted in principle. The above solution is generally applicable to a wide variety of machine-driven rotational processes with multiple and, in particular, periodically repeating revolutions, where automated and / or machine monitoring is to be carried out. For example, the rotary-driven object can be a machining tool used to process a workpiece, particularly for machining. Alternatively, it can be a driven spindle that, for example, rotates a machining tool or a workpiece. Furthermore, it can be a machine-driven object that is not directly related to workpiece machining. For example, it could be rotary-driven components of a pump, such as a pump shaft, a rotary piston, or an impeller. It could also be a rotary table and / or rotary-driven gripper systems of a rotary transfer machine. Other examples include turbines and industrial mills, agitators, or mixers. Acquiring a measurement with rotational angle resolution can be understood, in particular, as recording the values ​​of the measurement quantity as a function of the rotational angle. For example, each rotational angle can be assigned a corresponding value of the measurement quantity at that angle. The resolution and / or the interval between the rotational angles for which values ​​of the measurement quantity are recorded can depend, for example, on the device used to acquire the measurement quantity. In particular, the acquisition frequency or sampling rate and / or the resolution of this device can be relevant. Any rotation angle mentioned here can be selected from a range of 0° to 360° and / or can refer to a specific angular position within a single revolution, where the revolution comprises a rotational movement of the object from 0° to 360°. The rotation process can comprise several successive and / or contiguous revolutions, so that the object reaches and / or passes through the individual angular positions multiple times in succession. In particular, each angular position can be reached once per revolution and reached again in subsequent revolutions. In other words, each angular position can be traversed multiple times within the successive revolutions. Determining setpoint values ​​of the measured quantity with rotation angle resolution can be understood, in particular, as determining the setpoint values ​​as a function of the rotation angle. For example, at least one setpoint value can be assigned to each considered rotation angle, or a plurality of setpoint values ​​can be assigned within a range of values, see, for example, the tolerance band discussed below. The resolution and / or the spacing of the rotation angles for which setpoint values ​​are determined can, in turn, depend on the device used to acquire the measured quantity. In particular, the acquisition frequency and / or the resolution of this device can be relevant. Fundamentally, the rotation angle resolution of the setpoint values ​​and the measured quantity can be identical. The measured values ​​used to determine the target values ​​can be determined based on revolutions already completed during a specific rotation process. In particular, target value determination can be performed in real time and / or during the ongoing rotation process. The target values ​​can be determined exclusively or predominantly based on measured values ​​acquired during the rotation process. For example, no target values ​​may initially exist, or at most, preliminary target values ​​that are adjusted during the rotation process based on a number of subsequent revolutions. The target values ​​can be determined, at least partially, independently of historical measured values ​​or historical target values ​​derived from other, and especially past, rotation processes. Instead of relying solely on historical data for past rotation processes, target values ​​can be determined at least partially, or even exclusively, based on measured variable values ​​acquired during a currently executed rotation process. Specifically, measured variable values ​​from at least selected rotations completed up to the current point in time can be used to determine the target values. In particular, the target values ​​can be determined specifically for each rotational process. This can be advantageous in order to take into account batch-related influences of a workpiece currently being machined or environmental influences. The deviation can be determined based on a measured value obtained for a subsequent and / or current revolution. This is particularly relevant when continuing the rotation process after target values ​​have already been determined. For example, measured values ​​from completed revolutions can be used to determine the target values, and the subsequent measured value, whose deviation is to be determined, can be compared with at least one of these target values. This subsequent value can then be used to update the target value under consideration, and the updated target value can subsequently be used to determine the deviation for future revolutions. With the solution disclosed herein according to the independent claims, it is generally possible to determine deviations and any related errors during movement, and in particular during a rotational process, or, in other words, in real time. The solution is particularly distinguished by the fact that the deviations, and thus errors, can be determined based on target values ​​derived from measured values ​​acquired during the movement. In other words, a movement- or rotation-specific reference for determining errors can be established and, in particular, continuously updated during the movement or rotational process based on recorded measured values, namely in the form of target values. This allows, for example, better consideration of batch-dependent influences on a machined workpiece and / or eliminates the need to predefine imprecise limit values ​​for error determination. Furthermore, the solution is particularly suitable for motion and especially rotation processes for which no precise information on suitable limit values ​​of measured variables and / or target value deviations is available beforehand, on the basis of which errors could be detected, or for which such information would be very difficult to obtain. Because the solution disclosed here determines a process-specific and / or real-time reference for deviation analysis in the form of, for example, angle-resolved and, in particular, continuously updated target values, corresponding prior knowledge or preliminary investigations are not strictly necessary. Furthermore, the disclosed solution is characterized by the rotation angle resolutions explained above, or more generally, time and / or position resolution, so that errors can be detected quickly and, for example, immediately within the context of a currently executed motion cycle. In particular, the present solution can be used to monitor manufacturing processes that are based on a rotating or otherwise moving tool or other object, and in which process feedback is reflected, for example, in sensor-based measured variables. Specifically, recurring periodic patterns in the measured variables can be identified, where the period can correspond to a tool revolution or, more generally, a motion cycle. It has been found that, under constant and error-free process conditions, these patterns, and thus the measured variable values, can only differ slightly from revolution to revolution (or from motion cycle to motion cycle) for the same rotation angle, the same time within a motion cycle, and the same object position.The solution revealed here takes advantage of this by allowing previously recorded measurement values ​​to be used to determine the target values, without requiring precise process and / or workpiece knowledge or general reference values ​​for at least one measurement variable from the outset. In general, the setpoint can be an expected value or arithmetic mean of previously recorded measured values ​​for a considered angular position, a specific time, or a general movement position. One embodiment provides that the setpoints, e.g., those dependent on the rotation angle, are adjusted during the execution of further movement cycles and, in particular, are continuously adjusted. For example, a measured value recorded for each error-free revolution and for each time, movement position, and / or angular position can be used to update the setpoint for, e.g., this angular position. This can occur, in particular, before the same angular position, time, or movement position is reached again in a subsequent movement cycle. In this way, temporal changes during operation can be taken into account and / or a type of sliding setpoint determination can be performed.Such improved adaptation to current operating conditions can enhance the accuracy of deviation determination. Another embodiment provides for an additional check to determine whether the measured deviation meets at least one error criterion. The error criterion can be predefined and / or defined in relation to target values, particularly those specific to the process. For example, the error criterion can be defined based on a permissible relative or absolute deviation of a measured value from at least one target value at a given rotation angle, time, or position of movement, or it can specify a deviation. The check for compliance with the error criterion can be performed in real time, which contributes to the early detection of irregularities and proactive process monitoring. Another embodiment provides that the error criterion is based on an error ratio that relates the number of times or motion positions at which the measured quantity deviates impermissibly from the respective at least one target value to the number of times or motion positions within this motion cycle at which no such impermissible deviation exists. For example, the error ratio can relate the number of rotation angles, or in other words, rotation angle positions, at which the measured quantity deviates impermissibly from the respective at least one target value of the corresponding rotation angles or rotation angle positions to the number of rotation angles or rotation angle positions at which no such impermissible deviation exists.The permissible or impermissible deviation from the target value can be defined relative to the target value, which is determined specifically for each process, for example, as a percentage or another relative deviation. The error ratio can be determined individually for each completed motion cycle or continuously during the execution of a motion cycle. In particular, the error ratio can be determined on a motion cycle-specific basis. Errors can thus be detected early, especially in incomplete motion cycles, if the error ratio is continuously determined and, in particular, accumulated as described. Another embodiment provides that the error criterion is defined based on a statistical deviation parameter, which is calculated based on deviations of the time-resolved or position-resolved values ​​of the measured quantity within a motion cycle from the respective target values ​​of the times or motion positions within the motion cycle. For example, the error criterion is defined based on a statistical deviation parameter, which is calculated based on deviations of the rotation-angle-resolved values ​​of the measured quantity within one revolution from the respective target values ​​of the rotation angles. Possible parameters include, for example, the square root of the mean square error, the mean deviation, or the mean square error. In this way, a kind of summary analysis of the deviations within a respective motion cycle can be enabled, which compensates for smaller fluctuations.This can improve the accuracy of fault detection while still allowing for early detection. Another embodiment provides that, upon fulfillment of the error criterion, a warning message is issued to an operator of a device performing the rotation process, and / or a signal is issued that causes the device to stop the rotation process or, more generally, the movement. This signal can, for example, be a control signal that can be used directly by the device's drive devices to terminate the rotation process or other movement, and / or it can be a signal that the device can convert into such a control signal. In this way, deviation and error detection is enhanced by the possibility of timely corrective action, which improves operational reliability and process quality. Another embodiment provides that the measured variable is one of the following: a drive current or other drive parameter of a drive device, which, for example, generates a drive force for executing the rotation process or another movement; a position measurement of a moving unit, which, for example, performs the rotation process or another movement itself or moves an object performing the rotation process; a force value, which, for example, is dependent on the rotation process or another movement and corresponds, for example, to a reaction force from a machining process that includes the rotation process or another movement; a temperature value; a vibration value; or an acoustic quantity. The last three values ​​or quantities mentioned can vary, for example, depending on the rotation process or other movement and, in particular, on a machining operation carried out thereby. Another embodiment provides for determining the average drive current of a device performing the rotation or other movement. In particular, this embodiment further includes, depending on a change in the average drive current, at least one of the following measures: - at least temporarily deactivating the deviation detection and / or the error criterion verification described above; - adjusting the error criterion and, in particular, at least one limit value underlying this criterion and / or on the basis of which the presence of an error can be determined. For example, the limit value can be a permissible maximum deviation value. A changing average drive current can, for example, indicate changing process conditions and / or loads. In this case, the error criteria can be adjusted to avoid false alarms.For example, in the case of large changes in the drive current, higher average deviations from the setpoint may be allowed before an alarm is triggered. It has been recognized that the average drive current can be a criterion for the existence of discontinuous conditions under which the rotational process or other movement is performed. This discontinuous state can be detected and, in particular, appropriately compensated for using the aforementioned embodiment. Specifically, it has been recognized that under changing process conditions (e.g., altered radial / axial feed of a cutting tool performing the movement), the average cutting force and / or the average spindle current, for example, vary. In such a case, a systematic trend in the measured values ​​from one tool revolution to the next may be present. If the target values ​​were determined based on this, for example, according to the variants disclosed above, without taking the corresponding trend into account, this could impair the precision of the fault detection. Another embodiment provides for the partial estimation of at least selected target values, taking into account time-dependent changes in the measured variable, particularly using an autoregressive model or another model for detecting time-dependent patterns. For example, the target values ​​can be determined partly based on a moving average or expected value calculation, and partly based on a time-dependent and / or statistically determined component. This latter component can be configured, in particular, to account for a time-dependent trend in the measured variable. In this way, temporal patterns and trends can be at least partially considered, especially to at least partially compensate for the previously described influences of discontinuous conditions.In particular, in this case, monitoring of the rotation process can continue continuously and / or adjustments to the error criterion can be omitted despite changing conditions. Alternative approaches to using an autoregressive model can include, for example, exponential smoothing, Bayesian methods, and linear or nonlinear regression models. The invention also relates to an arrangement for detecting deviations in a device performing a movement, in particular a machining or forming workpiece processing device, wherein the device has a drive device for driving an object and thereby performing the movement, which has several successive movement cycles, wherein the arrangement comprises: - a detection device for detecting at least one time-resolved or position-resolved measured quantity that is variable during the movement; wherein the arrangement is configured to: - determine time-resolved or position-resolved target values ​​of the measured quantity, wherein at least one target value for the respective time or movement position is determined on the basis of a plurality of values ​​of the measured quantity detected for the same time or the same movement position within the movement cycles;and to record, for at least one point in time or movement position, a deviation of at least one further recorded value of the measured quantity from at least one target value for that point in time or movement position. As a specific example, an arrangement for deviation detection in a device performing a rotational process, in particular a machining or forming workpiece processing device, is disclosed, wherein the device has a drive device for rotating an object and thereby performing the rotational process, which has several successive revolutions, wherein the arrangement comprises: - a detection device for detecting at least one rotationally angle-resolved measured quantity that is variable during the rotational process; wherein the arrangement is configured to: - determine rotationally angle-resolved target values ​​of the measured quantity, wherein at least one target value for the respective rotational angle is determined based on a plurality of values ​​of the measured quantity detected for a respective rotational angle;and - to detect, for at least one rotation angle, a deviation of at least one further recorded value of the measured quantity from at least one target value for that rotation angle. The arrangement according to any of the foregoing variants can be configured to carry out a method according to any embodiment disclosed herein. The arrangement can include any further features necessary to carry out all of the method steps disclosed herein. According to a further aspect of the invention, a system is also disclosed that comprises an arrangement for detecting deviations as well as a device for performing and / or driving the rotation process according to any of the examples disclosed herein. The invention also relates to a method for determining deviations during machining or forming of a workpiece, the method comprising: - capturing at least one time-resolved or position-resolved measured quantity that is variable during a movement performed in the context of the workpiece machining, wherein the movement comprises several successive movement cycles; - determining time-resolved or position-resolved target values ​​of the measured quantity, wherein, based on values ​​of the measured quantity that are each captured for the same time or the same movement position within the movement cycles, at least one target value for the respective time or movement position is determined; - determining, for at least one time or movement position, a deviation of at least one further captured value of the measured quantity from at least one target value for this time or movement position. According to a specific example, a method for determining deviations during forming or machining of a workpiece is disclosed, wherein the method comprises: - Acquiring at least one rotation-angle-resolved measured quantity that is variable during a rotation process carried out in the context of machining the workpiece, wherein the rotation process comprises several successive revolutions; - Determining rotation-angle-resolved target values ​​of the measured quantity, wherein at least one target value for the respective rotation angle is determined based on a plurality of values ​​of the measured quantity acquired for a respective rotation angle; - Determining, for at least one rotation angle, a deviation of at least one further acquired value of the measured quantity from at least one target value for this rotation angle. The invention also relates to an arrangement for deviation detection in a machining or forming workpiece processing device, wherein the workpiece processing device has a drive device for driving an object and thereby performing a movement carried out in the context of machining the workpiece, which comprises several successive revolutions; wherein the arrangement has a detection device for detecting at least one time-resolved or position-resolved measured quantity that is variable during the movement; wherein the arrangement is configured to: - determine time-resolved or position-resolved target values ​​of the measured quantity, wherein at least one target value for the respective time or position is determined based on a plurality of values ​​of the measured quantity detected for the same time or position within the movement cycles;and - to record, for at least one time point or movement position, a deviation of at least one further recorded value of the measured variable from at least one target value for that time point or movement position. The arrangement may include any further features necessary to carry out all the procedural steps disclosed herein. According to a further aspect of the invention, a system is also disclosed that comprises an arrangement for detecting deviations in a machining or forming workpiece processing device as well as a drive device and / or a workpiece processing device according to any example disclosed herein. Exemplary embodiments of the invention are explained below with reference to exemplary configurations. Fig. 1 shows an arrangement for determining deviations in a machining device; Fig. 2 shows, in partial views a) and b), exemplary values ​​of a measured quantity obtained over several tool revolutions; Fig. 3 shows a comparison of the measured quantity values ​​with determined target values ​​or a tolerance band derived therefrom; Fig. 4 shows, in partial views a)-c), further examples of obtained values ​​of a measured quantity and error characteristics derived therefrom. Fig. 1 shows an arrangement 10 for determining deviations, which is installed in a workpiece machining device 11. The arrangement 10 and the workpiece machining device 11 form a system 1 according to an embodiment of the invention. The workpiece machining device 11 is an exemplary device 13 that performs a rotation process monitored according to the invention. The workpiece machining device 11, which is only partially shown and can be configured according to known examples of machine tools, is set up to perform machining operations on the workpiece. It comprises a drive device 12 in the form of a spindle. The drive device 12 is configured to drive a tool holder 14 together with a milling head 16 inserted therein, so that the milling head 16 performs a multitude of successive revolutions about its longitudinal axis L. The spindle is guided along a workpiece 18 in a feed direction V by means of a feed device (not shown separately), or, in other words, a feed axis. The workpiece 18 is positioned on a machining table 20. The arrangement 10 for deviation determination comprises at least one detection device 22 for acquiring values ​​of a predetermined measured quantity. The detection device 22 can, for example, be a detection device 22 already present in the workpiece machining device 11, or one that is to be provided for in any case, or it can be selectively retrofitted as part of the arrangement 10. In the example shown, the detection device 22 detects a drive current of the drive device 12, which can also be referred to as the spindle current. This detection is performed with rotation angle resolution, for which the detection device 22 can, for example, access measurement data from a rotary encoder (not shown) that detects the rotation angle of the drive device 12, and / or control data from a control unit 21 of the workpiece machining device 11. This control data can control the drive device 12 to execute the rotational movement.According to one embodiment, the measured quantity is acquired as a function of time, and the controller receives information about the rotation angle of the drive device as a function of time via the described rotary encoder. A rotation angle-resolved representation of the measured quantities is possible over time. As is particularly clear from the general description, a conversion and thus resolution in angles is not mandatory; time-resolved measured values ​​could also be considered. In this case, signals from the rotary encoder can at least be taken into account to the extent that completed rotations and thus completed motion cycles are recognized. The measured values ​​can each be assigned to specific points in time within individual motion cycles. These points in time can each be defined relative to a common reference value, for example, relative to the start of a respective motion cycle or a specific cycle duration. In this way, identical points in time within the motion cycles can be defined and compared, e.g., mean points in time or points in time that are a certain fraction of the cycle duration after the start of a respective motion cycle. The arrangement 10 also includes a detection device 24, which could optionally be integrated into the control device 21 or the recording device 22, for example as a software module. Possible data connections between the aforementioned devices 21, 22, and 24 are shown by dashed lines in Fig. 1. The determination unit 24 is designed to determine target values ​​resolved by rotation angle based on the measured values ​​determined by the acquisition unit 22, as well as deviations of these target values ​​from measured values ​​acquired during further processing. This will be explained in more detail below. Furthermore, the detection device 24 is configured to determine whether a predefined error criterion has been met, as illustrated below. If the criterion is met, the detection device 24 can, for example, send a signal to the control device 21 via a direct signal connection (not shown in Fig. 1) (or, for example, indirectly via the detection device 22) to trigger warning messages, an emergency stop, or the like. In general, the embodiments disclosed here are particularly suited for monitoring manufacturing processes, such as, for example, a machining process in which the process comprises a motor-driven rotary motion (e.g., milling spindle) of a tool (e.g., drill or milling cutter) and a mechanical feedback effect from the process to the tool exists (e.g., cutting force). The rotary motion is an example of a rotational process within the meaning of the claims, wherein the tool is an example of an object within the meaning of the claims. Information from the manufacturing process, and in particular measured variables whose deviations from target values ​​are to be determined for the purpose of fault detection, can be acquired, for example, via sensors on the mechanics (e.g., positions of feed axes) or on the motor (e.g., motor current). In contrast to Fig. 1, one could, for example, consider a press that forms, repeatedly punches, or cuts a workpiece. The press can perform repeated motion cycles comprising linear stroke movements. The following error analysis methods can be applied analogously to such an example, and in particular to linear motion positions instead of rotational angles. Fig. 2 shows an exemplary recording of measured values ​​that can be acquired by the sensing device 22 over a certain period of time, with the period plotted along the respective horizontal axis. In partial view a), an actual x-position in millimeters is shown along the vertical axis. The x-position corresponds to the position along a machine axis of the workpiece machining device 11 from Fig. 1 and, in the example shown, to the feed axis, which is configured to execute a movement in the feed direction V. In the part view b), the actual z-position is shown in millimeters, which corresponds to a position along an optional additional machine axis, for example, running orthogonally to the feed axis. An x / z origin position along the range of motion of the respective machine axes can be defined according to common examples of the prior art. The position values ​​can also be determined using known approaches of the prior art, for example from corresponding detection devices and / or control signals and / or drive currents for the respective machine axes. Figures 2a) and 2b) each show two exemplary measurement curves, A and B. Measurement curve A relates to a milling tool with new cutting edges. Measurement curve B, on the other hand, relates to a tool with a broken cutting edge. Figures 2 and 3 show that the recorded measurements repeat periodically according to a specific pattern. This pattern occurs within a single tool revolution, as indicated. The solution disclosed here takes advantage of this by using such patterns as a reference for a currently executed rotation process in order to identify abnormal deviations for future tool revolutions, which may indicate errors. More precisely, continuously measured value profiles of a considered measurement quantity, as shown in Figure 2, can be segmented, with each segment corresponding to a single tool revolution. Furthermore, each segment is subdivided into the individual rotation angles traversed during one revolution.For example, these angular positions can be calculated based on the rotational speed and segment duration, and / or the instantaneous rotational speed and time, and / or they can be recorded from the outset using a rotary encoder. Therefore, for each segment, a measured value is available for every considered rotational angle. As mentioned, the detection device 22 from Fig. 1 can, according to exemplary embodiments, be configured to detect the measured values ​​from Fig. 2 and assign them to a current rotation angle, e.g., via a flag, a data array or database entry, or similar. In the example considered here, the detection device 22 is additionally or alternatively configured to detect the spindle current of the feed axis as a measured variable. Although this is not shown in a separate figure, it also exhibits a repeating pattern over several successive revolutions, as can be seen for the measured variables in Fig. 2. With reference to Fig. 3, an exemplary setpoint determination using the measuring device 24 is explained below. If the tool 16 has already completed several revolutions under constant process conditions, for example, a constant feed rate, the rotation angle-resolved measured values ​​of each revolution can be averaged, for example, to determine a setpoint for a respective rotation angle. Alternatively, the setpoint can be determined, for example, as an expected value, particularly assuming a normal distribution or another statistical distribution. More precisely, after N revolutions, N measured values ​​are available for each rotation angle within a degree of freedom, which ranges, for example, from 0° to 360°. From these, the mean or expected value can be determined to establish a setpoint for the measured variable for a respective rotation angle as a future expected value.to determine a given rotation angle position. Alternatively, an autoregressive model can be formed based on the N measured values ​​of each rotation angle to determine the expected value and, optionally, a standard deviation for the next expected measured value at that rotation angle. For a constant rotational speed, the sampling times of the sensor data and the corresponding rotation angles can be directly considered across all revolutions. For a varying rotational speed, an interpolation algorithm can be used to obtain N measured values ​​for the same rotation angles for all revolutions. Figure 3 illustrates this for a single tool revolution, using the spindle current in amperes as the measured variable. The measurement curve was recorded for a tool with five cutting edges, so ideally, five similar deflections should be visible. However, small runout errors between cutting edges (or slightly different tool radii) result in less periodic deviations. Fig. 3 shows a tolerance band determined based on the target values ​​recorded for completed revolutions of a current machining operation. For this purpose, a target-value-specific tolerance range is added to the rotation-angle-resolved target values. This tolerance range can, for example, encompass one or more times the standard deviation of the respective measured values ​​for a given rotation angle. This tolerance range includes values ​​both above and below the target value and can therefore be described as a + / - value range. The respective target value can, for example, be a zero value or a mean value within this range. In the example shown, the tolerance range is determined based on the standard deviation of the N measured values ​​for a given rotation angle. The tolerance band or tolerance range can be understood as a target value range and can specify several permissible target values ​​for each rotation angle. Another curve shows the so-called next measured values, which are determined for a revolution following the revolution whose measured values ​​were last used to determine the target values ​​or the displayed tolerance band. It is shown that these measured values ​​lie completely within the tolerance band, so the revolution is recognized as error-free. Consequently, these measured values ​​are also used for an immediate update and / or a rolling determination of the tolerance band in real time. To limit the computational effort, the measured values ​​considered for tolerance band determination can be limited to a specific number of already completed revolutions (for example, no more than ten or no more than twenty), in particular to a corresponding number of the most recent and / or current revolutions. Figure 3 also shows a case where the measured values ​​for one revolution are at least partially outside the tolerance band. This can indicate a process error, such as cutting edge breakage, and can be detected accordingly. In principle, exceeding the tolerance band for a single rotation angle can be defined as an error criterion. However, to increase the detection accuracy of process errors and, in particular, to avoid misdiagnoses, error criteria and error detection strategies according to further embodiments of the invention are discussed with reference to Fig. 4. These can, for example, at least partially suppress the influence of signal noise and / or material inhomogeneities in the workpiece in order to avoid falsely diagnosing a cutting edge breakage or other defects. In partial view a) of Fig. 4, exemplary measured values ​​in the form of a spindle current A are recorded over a multitude of revolutions; see the correspondingly long covered period along the horizontal axis. This period corresponds to a workpiece machining operation, and more precisely, a milling operation of a workpiece, performed by the workpiece machining device 11 from Fig. 1. Material is removed from the workpiece under largely constant process conditions. Simultaneously, three bores within the workpiece are traversed by the milling tool. At the beginning and end of the depicted period, the milling tool engages and retracts from the workpiece, respectively, which is why the measured values ​​exhibit a largely aperiodic pattern typical for these processes. Between engagement and retraction, the measured values ​​fluctuate largely periodically, with the exception of three breakout areas marked A. These breakout areas result from the milling tool passing over the aforementioned bores. Fig. 4 shows regular measured values ​​recorded by the acquisition device 22 from Fig. 1 during error-free machining. These can be used, for example, as training or reference data by the determination device 24 from Fig. 1 to determine target values ​​and, if necessary, a tolerance band of the type described above. Also shown are abnormal measured values ​​recorded during defective machining, for example, in the event of a cutting edge breakage of the milling tool 16 from Fig. 1. For both of the regular and abnormal measured variable values, the exemplary error parameters shown in partial views b) and c) of Fig. 4 are determined by the measuring device 24 from Fig. 1. Referring first to partial view b), the determination of an error ratio rV is explained. The error ratio rV is determined individually for each revolution and is a ratio of a first number of rotation angles and a second number of rotation angles. The first number of rotation angles indicates the number of rotation angles within the respective revolution at which the measured quantity deviates impermissibly from at least one target value of a respective rotation angle, for example, because it lies outside a tolerance band defined analogously to Fig. 3. The second number of rotation angles indicates the number of rotation angles within the same revolution at which no such impermissible deviation exists. For a substantially error-free revolution, the error ratio rV has a value equal to or close to zero. This is shown in partial view b) of Fig. 4 by the lower measurement curve. This measurement curve comprises the values ​​of the error ratio rV that are determined for each revolution performed during the milling operation described in detail view a). Again, the depicted time course can be clearly segmented into individual revolutions and / or rotation angles in the manner described above. More precisely, this measurement curve shows the revolution-specific error ratios rV, which are calculated from the regular measured values ​​from partial view a) in Fig. 4. The upper measurement curve in detail view b) of Fig. 4, however, relates to a defective machining process according to the abnormal measured values ​​from detail view a). It can be seen that in this case the values ​​of the rotation-specific error ratio rV are significantly above 0 and partly close to 1. Accordingly, an error criterion can be defined as (and / or an error can be detected if) the error ratios rV of the respective executed revolutions exceed a permissible value and / or the error ratio rV between two successive revolutions increases beyond a permissible level. For the sake of completeness, it should be noted that the traversing of the aforementioned workpiece bores, along with the associated effect on the spindle current, is an example of discontinuous process conditions. Compensation strategies to nevertheless ensure reliable error detection and, in particular, to account for any trend developments in the measured values, are explained in the general description section and can also be optionally implemented in the present embodiment. Referring to partial view c) from Fig. 4, the statistical deviation parameter "root mean square error WMQFd" (RMS) is shown below as an alternative or additional error parameter. This parameter is determined for both error-free (i.e., regular) and error-prone (i.e., abnormal) measured variables as shown in partial view a) from Fig. 4. The error parameter WMQFd is calculated on a revolution-specific basis, based on deviations of the rotation-angle-resolved values ​​of the measured variable within each revolution from the respective target values ​​of the rotation angles. According to the known method of calculation for this error parameter, the difference between each actual measured variable value and each predicted or target measured variable value is calculated for each rotation angle within a revolution. These differences are then squared.The square root of the average of these squared deviations is then calculated. For the lower measurement curve of the partial view c) of Fig. 4, which concerns regular or error-free operation, the revolution-specific error characteristics WMQFd during machining are again exactly or close to 0. For the upper measurement curve of the partial view c) of Fig. 4, which concerns the abnormal or faulty operation, the rotation-specific error characteristics WMQFd during machining are sometimes up to ten times and in any case predominantly significantly more than 1. As above, the error criterion can be defined, for example, as exceeding a permissible maximum value of the error indicator WMQFd or a permissible jump height of the error indicator WMQFd between successive revolutions. The aforementioned error parameters rv and rd can be calculated once per tool revolution. This saves computation time but reduces the time intervals at which an anomaly can be detected (i.e., detection is only possible after each fully completed tool revolution). Alternatively, the error parameter values ​​can be recalculated and / or updated with each newly sampled measurement. In this case, very fast anomaly detection is possible even during an ongoing revolution. QUOTES INCLUDED IN THE DESCRIPTION This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature DE 10 2016 100 503 B3

[0011] Zitierte Nicht-Patentliteratur Zhou & Xue: Review of tool condition monitoring methods in milling processes, Int. J. Adv.Manuf. Technol. 96 (2018), 2509-2523

[0006] T. Mohanraj: Tool condition monitoring techniques in milling process - a review, J. Mater. Res. Technol. 9 (2020),1032-1042

[0006] Li X.: Detection of Tool Flute Breakage in End Milling Using Feed-Motor Current Signatures, IEEE / ASME Trans. Mechatronics. (2001) 6:491-243

[0008] Altintas Y., Aslan D.: Integration of virtual and on-line machining process control and monitoring, CIRP Ann. (2017) 66:349-352

[0010]

Claims

Method for determining deviations in a movement, which is carried out in particular during machining or forming of a workpiece and / or machining of a workpiece using a press, comprising: - recording at least one time-resolved or position-resolved measured quantity that is variable during the movement, wherein the movement is performed by a driven object (16) and comprises several successive movement cycles; - determining time-resolved or position-resolved target values ​​of the measured quantity, wherein at least one target value for the respective time or movement position is determined from a plurality of values ​​of the measured quantity, each of which is recorded for the same time or movement position within the movement cycles;- Determine, for at least one point in time or movement position, a deviation of at least one other recorded value of the measured quantity from at least one target value for that point in time or movement position. Method according to one of the preceding claims, wherein a respective time point within a motion cycle is defined relative to a reference time point and / or a period of motion cycle; or wherein the motion process is a rotation process and the motion position is a rotation angle; or wherein the motion process is axial and in particular linear and wherein the motion position is a position along the axis of motion. Method according to one of the preceding claims, wherein the setpoints are adjusted when further motion cycles are performed. Method according to one of the preceding claims, further comprising: checking whether the determined deviation fulfills at least one error criterion. Method according to claim 4, wherein the error criterion is defined based on an error ratio (rv) that relates the number of time points or movement positions at which the measured quantity deviates impermissibly from the respective at least one target value to the number of time points or movement positions of this movement cycle at which no such impermissible deviation exists. Method according to claim 3 or 4, wherein the error criterion is defined based on a statistical deviation parameter (WMQFd) calculated on the basis of deviations of the time-resolved or position-resolved values ​​of the measured quantity within a motion cycle from the respective target values ​​of the times or motion positions within the motion cycle. Method according to one of the preceding claims, wherein the measured quantity is one of: - a drive current or other drive quantity of a drive device (12); - a position measurement of a moving unit, in particular a machine axis; - a force value; - a temperature value; - a vibration value; - an acoustic quantity. A method according to one of the preceding claims with claim 4, wherein the method further comprises: - determining an average drive current from a drive device (12) performing the movement; and depending on a change in the average drive current further comprises at least one of the following measures: - at least temporarily deactivating the deviation detection and / or the error criterion verification; - adjusting the error criterion, in particular at least one limit value underlying this error criterion. Method according to one of the preceding claims, further comprising: - Estimation of at least selected of the target values ​​taking into account time-dependent changes in the measured variable, in particular by means of an autoregressive model or another model for detecting time-dependent patterns. Arrangement (10) for detecting deviations in a device (13) performing a movement, in particular a machining or forming workpiece processing device (11) and / or workpiece processing by means of a press, wherein the device (13) has a drive device (12) for driving an object and thereby performing the movement, which has several successive movement cycles, wherein the arrangement (10) comprises: - a detection device (22) for detecting at least one time-resolved or position-resolved measured quantity that is variable during the movement;wherein the arrangement (10) is configured to: - determine time-resolved or position-resolved setpoints of the measured quantity, whereby at least one setpoint for the respective time or movement position is determined based on a plurality of values ​​of the measured quantity recorded for the same time or movement position within the movement cycles; and - for at least one time or movement position, detect a deviation of at least one further recorded value of the measured quantity from at least one setpoint for that time or movement position. Method for determining deviations during machining or forming of a workpiece and / or machining of a workpiece using a press, comprising the method of: - Acquiring at least one time-resolved or position-resolved measured quantity that is variable during a movement performed in the context of the workpiece machining, wherein the movement comprises several successive motion cycles; - Determining time-resolved or position-resolved target values ​​of the measured quantity, wherein, based on values ​​of the measured quantity that are each acquired for the same time or position of movement within the motion cycles, at least one target value for the respective time or position of movement is determined; - Determining, for at least one time or position of movement, a deviation of at least one further acquired value of the measured quantity from at least one target value for that time or position of movement.System (1) comprising a machining or forming workpiece processing device (11) and an arrangement (10) for deviation detection, wherein the workpiece processing device (11) has a drive device (12) for driving a tool (16) and thereby performing a movement carried out in the context of workpiece processing, comprising several successive motion cycles; wherein the arrangement (10) has a detection device (22) for detecting at least one time-resolved or position-resolved measured quantity that is variable during the rotation process;and wherein the arrangement (10) is configured to: - determine time-resolved or position-resolved setpoints of the measured quantity, wherein, based on a plurality of values ​​of the measured quantity, each recorded for the same time or position of movement within the movement cycles, at least one setpoint for the respective time or position of movement is determined; and - for at least one time or position of movement, to detect a deviation of at least one further recorded value of the measured quantity from at least one setpoint for that time or position of movement.

Images