Method and system for improving the actual accuracy of an output-side movement

Abstract

The invention relates to a method and a system for improving the actual accuracy of an output-side movement of an electric drive (2) having a connected mechanical transmission device (6). The axes (12) of a machine are controlled by at least one electronic control device (3). A predetermined target specification is followed by system-related actual deviations. The occurring output-side movement is detected with the aid of a measuring system (15), in particular a secondary sensor (10), connected to the mechanical transmission device (6), and made available to a self-learning controller (14). Parameters (16) and / or signals (17) are generated by the self-learning controller (14). The introduction, in particular by feedforward control, of the generated parameters (16) or signals (17) into the system leads to the at least partial improvement of the actual accuracy of the output-side movement. The parameters (16) and / or signals (17) can be used both in the system in which the learning has taken place and in systems that have sufficient similarities to the learning system in order to increase the accuracy.

Description

[0001] Method and system for improving the actual accuracy of an output-side movement

[0002] The invention relates to a method for improving the actual accuracy of an output-side movement of a system for generating machine movements and to a system designed for this purpose, as specified in the claims.

[0003] The invention is designed to optimize the actual accuracy of a drive, in particular of an industrial robot or other machine. High actual accuracy is to be understood as the smallest possible deviation between the desired target profile and the practically achievable actual profile, e.g. of a robot path. In the context of the invention, actual accuracy is primarily understood to mean movement accuracy. Furthermore, it includes the temporal course of the position accuracy or its temporal changes (speed, acceleration) of the machine. The axes of this single- or multi-axis machine are controlled by at least one electronic control device in such a way that a predetermined target specification is or can be executed, typically by the so-called TCP (tool center point) or end effector of the machine, with the smallest possible or negligible actual deviations.If the desired trajectory is as precise as possible, this is particularly useful for two- or three-dimensional trajectories. This applies especially to machines with coupled or serially arranged axes.

[0004] It has been recognized that in applications which place the highest demands on actual accuracy, the properties of the mechanical drive train can also be responsible for deviations. In particular, the mechanical transmission system from the driving motor to the joint or to the linear guide system, for example comprising gear drives, toothed belts or chain drives, is often responsible for positioning or movement deviations. In many cases, the measuring system for the position of a movement axis is a displacement sensor which is mounted on the motor side, i.e. before the mechanical transmission system. As a result, deviations from ideal behavior which are caused by the mechanical transmission system cannot be recorded and impair the accuracy with which the joint or linear axes can be moved. In particular, these are deviations from the linear transmission behavior (ora constant gear ratio) and variable elasticities and friction. These arise from the design of gears or toothed belts in the drive train and generate specific deviation patterns for each path movement.

[0005] For example, one of the influencing factors is often simplified as gear backlash, which describes a difference between the target and actual position that depends on the direction of movement or the direction of the load torque. In more precise modeling, a hysteresis between the position on the motor side and the position on the output side is often used. During the execution of a movement, additional non-uniform deviations can occur. Even with an exactly constant speed of the drive motor, non-uniform movement can occur on the output side. This results mainly from the non-constant gear ratio in interaction with friction, elasticity and inertia in the drive train and the load. These deviations are very difficult to compensate for with high accuracy because the effects are very complex and depend on many parameters such as position, speed, load, etc.At the same time, however, these deviations are the ones that significantly limit the use of robots for high-precision applications.

[0006] To measure the deviations that occur in the drive train, it is a well-known procedure to install a position or angle measuring system on the output side of the axis. Such measuring systems are often referred to as secondary encoders, in contrast to motor-side, primary encoders or position sensors. This design is frequently used in machine tools, for example, but also in robots. Furthermore, compact drive units are often used in robotics and special-purpose mechanical engineering. These combine a motor with a gearbox and often also a drive controller, which also offer an output-side position sensor as an option.

[0007] Document US5155423A describes a robot in which the position of the robot's links is detected at the joint end—i.e., at the end of the drive train. Document US6258007B1 describes a drive module consisting of a motor and gearbox, suitable for robot construction, which has both a motor-side and an output-side position sensor.

[0008] The document US20100191374A1 describes a method for feedback of signals from secondary encoders on a robot, in which a higher bandwidth is achieved by a special design of the calculated feedback signals.

[0009] State-of-the-art technology also allows for the reduction of machine axis deviations—such as gear backlash or hysteresis—by integrating the output-side sensor into an instantaneous control loop. A common approach here, for example, is to use a position sensor on the motor side for the speed control loop and a position sensor on the output side for the position control loop. A mix or combination of the position sensor signals in feedback control loops is also common. In common procedures, the control correction is always determined directly from the secondary measuring system or with its assistance, in accordance with the principle of a feedback control loop.

[0010] However, when the highest accuracy is required during movement, such control loops for compensating deviations have certain disadvantages: Some of the deviations are caused by deviations in the transmission behavior of the drive trains, for example due to the teeth in the gearbox or a toothed belt. This leads to relatively high-frequency disturbances or deviations that may be in the frequency range of the natural frequencies of the mechanical system or even higher, which makes stable compensation difficult. As is known from control engineering, for stable operation of control loops the gain in the loop must be less than 1 for phase shifts greater than 90°. This can only be achieved by limiting the bandwidth, or certain improvements can be achieved by using a frequency response correction that takes the natural frequencies of the mechanical system into account.Overall, the achievable frequency range for compensation, and thus also the achievable improvement in accuracy, remains limited in this type of control. Iterative methods, which are used for repeated processes, are known as alternative methods for improving the accuracy of drive systems or robots. Here, compensation is not determined spontaneously or immediately in a feedback loop, but rather an attempt is made to improve a compensation signal from one iteration to the next. When the process is repeated, the deviations are measured, and these deviations are used to determine a correction value for the compensation signal. This compensation signal is stored in a memory and is available for further iterations or can be continuously improved over several iterations. These methods are known as "iterative learning control" or "repetitive learning control."

[0011] The publications US2009 / 0222109A1, DE102011011679A1 and EP1647369A2 describe learning compensation methods for improving the accuracy of robots or other machines.

[0012] Iterative, particularly learning-based compensation methods, have so far been applied using motor-side, i.e., primary-side, position sensors. However, this can only reduce the control deviation between the target and actual position of the motor. To capture the deviations described at the beginning due to effects in the drivetrain, it is important to perform a measurement on the load side.

[0013] In the learning compensation methods for compensating for deviations caused by the drive trains, measuring systems can be used that measure the position or movement of the tool or end effector virtually from a distance – that is, measuring systems that record quantities in Cartesian space. These include, for example, optical measuring systems such as laser trackers. However, these highly complex measuring systems often require special integration into the control system. Furthermore, they lead to certain other limitations or restrictive requirements in the application.

[0014] Stationary laser trackers enable very high-quality measurements, but they are also very expensive. Furthermore, these measurement systems require a relatively large amount of space, which is a significant disadvantage in confined production environments. Furthermore, an optical measurement system such as a laser tracker requires a constant line of sight between the stationary measuring device and the tracked measurement point. This is a significant limitation for many applications, especially if the end effector also needs to be reoriented.

[0015] Inertial sensors or acceleration sensors are relatively space-saving and inexpensive, but can only directly detect movement; they cannot measure position. Furthermore, such sensors usually require subsequent installation on a robot, which is typically relatively complex.

[0016] The object of the present invention was to overcome disadvantages of the prior art and to provide an improved device or an improved method for increasing the actual accuracy of an output-side movement of a system for generating repetitive or at least approximately repetitive machine movements.

[0017] This object is achieved by a method and a device according to the claims.

[0018] In the following, embodiments of the invention are presented, whereby it is pointed out that the embodiments can generally relate to both the method and the implementation form as a system.

[0019] This specification describes a method and a system that includes a self-learning controller for generating compensation data, in particular corrective parameters or signals. The generated compensation data is used at a later time to improve the actual accuracy.

[0020] Furthermore, “the measuring system connected to the drive train or transmission device” is sometimes also referred to as a “secondary measuring system,” “secondary sensor,” or “secondary encoder.” The “parameters or signals for improving the actual accuracy of the output-side movement” are sometimes also referred to as “compensation data” or “compensation signals.” Furthermore, the “setpoint signal” or “specified signal” used to control the drive controller is sometimes also referred to as a “comparable setpoint signal” or “comparable specified signal” to emphasize that it is a section of the setpoint signal that already exists in a similar or identical form as a setpoint signal. In some cases, the terms “setpoint signal” or “specified signal” can also be understood as “comparable setpoint signal” or “comparable specified signal.” The respective possibilities arise from the context.

[0021] In the context of this description, a "further system", or a "further technical system", or a "technically similar or comparable system" is understood to mean a system that is kinematically and dynamically identical or similar to the system in which or with which the corrective parameters or signals for improving the actual accuracy of the output-side movement were generated in advance.

[0022] The method according to the invention serves to improve the actual accuracy of an output-side movement of a system for generating machine movements, in particular repetitive or at least approximately repetitive, predetermined machine movements. In particular, the actual accuracy of an output-side movement of the system, for example, of at least one axis of a machine or a robot, is thereby improved with regard to time, speed, and / or position. The system comprises at least one electric drive with a mechanical transmission device connected to it. The method comprises the steps:

[0023] Controlling a drive controller of the electric drive in a first pass with a target signal or preset signal, whereby a drive signal is provided to the electric drive and whereby an output-side movement is generated at the output of the transmission device,

[0024] Detecting the output-side movement with the aid of a measuring system directly connected to the transmission device, i.e. mechanically or physically attached to the transmission device, in particular by means of a secondary sensor outside or away from the electric drive, providing the target signal and the detected movement to a self-learning control system,

[0025] Generating, in particular calculating, parameters or signals to improve the actual accuracy of the output-side movement by the self-learning control system based on the provided target signal and on the detected movement, in particular taking into account detected deviations, before the predetermined machine movement is repeated in at least one further run,

[0026] Introducing these pre-generated parameters or signals (i) by the self-learning controller (i) into the setpoint signal or into a setpoint signal of another technically similar system, and / or (ii) into the drive controller or into a drive controller of another technically similar system, and / or (iii) into the drive signal for the electric drive or into a drive signal for an electric drive of another technically similar system, and

[0027] Carrying out at least one further run of the predetermined machine movement based on the drive signal for the electric drive which has been changed, in particular influenced or modified, in advance by the generated parameters or signals.

[0028] In the context of the claimed invention, it is considered to be included that the corrective parameters or signals generated by the self-learning controller can alternatively also flow into an electronic controller or into a control device for the drive controller, with which the target signal or specification signal is provided.

[0029] It is essential that the output-side movement is recorded by a secondary measuring system located directly on the mechanical transmission device. This secondary measuring system measures the movement in the transmission path downstream of the electric drive. It is also essential that the recorded movement and the target signal or preset signal are fed to a self-learning controller, which generates parameters or signals in advance, particularly compensation signals, which can be used to subsequently improve the actual accuracy of the output-side movement.

[0030] It is therefore also important that the improvement in actual accuracy does not occur instantaneously or directly, but only in a later phase, namely when the drive controller of the electric drive is repeatedly activated with a comparable or modified setpoint signal or a comparable or modified specification signal. Typically, the output-side movements required in an application are repeated frequently. This can be the case, for example, when machining the next workpiece. In this case, the parameters or signals generated during the last run, which can be used to improve the actual accuracy of the output-side movement, can be used in the correspondingly repeated movement. This procedure improves the actual accuracy in the repeated movement.Even if a movement is repeated several times and already has improved actual accuracy, the corresponding recorded movement and the target signal or preset signal can be fed back to the self-learning controller and used to generate further improved parameters or signals. In this way, the movement can be further optimized iteratively with each iteration.

[0031] The integration of the generated parameters or signals into the comparable target signal is not necessarily limited to the same system, but can also be integrated into another technically similar system. If the two systems are sufficiently similar, improved actual accuracy is achieved in the technically similar system.

[0032] The repeated movement does not necessarily have to correspond to the next machine cycle or workpiece, but could also only comprise partial sections of a desired output-side movement.

[0033] The repeated movement does not have to be the result of the exact same setpoint signal, but can also be the result of a comparable setpoint signal or comparable specification signal. In this description, a comparable setpoint signal or comparable specification signal is understood to mean a setpoint signal or specification signal for which the actual accuracy of the movement is further improved by using parameters or signals that can be used to improve the actual accuracy of the output-side movement.

[0034] The measuring system is directly connected to the mechanical transmission device, in particular to its output-side output. The measuring system or its secondary encoder on the mechanical transmission device can be formed by angle or rotary encoders, incremental encoders, or absolute position encoders, and other measuring systems known from the prior art, or can include such components. The measuring system or its secondary encoder can also be formed by inertial measuring systems, such as acceleration sensors, gyroscopes, torque sensors, or tachogenerators, or can include such components.

[0035] The invention combines the measuring systems assigned to individual motion axes on the output side, known in the prior art for improving accuracy, with the method of iterative learning methods for determining compensation data or compensation signals. The integration of secondary measuring systems into driven machine axes is known per se, but their use in conjunction with iterative learning compensation methods is not.

[0036] Iterative learning processes and controls offer several advantages: The calculation of the correction parameters or the compensation or improvement signals takes place virtually offline, once the movement and measurement sequence is over. This allows any delay between the drive's setpoints and the measurement by the secondary encoder to be taken into account in the next run. Furthermore, stabilizing filters can be used that do not cause a phase shift, enabling a higher stable bandwidth. Furthermore, the corrections, i.e. the corrective parameters or signals, can be determined over multiple learning cycles. This contributes both to stability and to a better end result in nonlinear physical relationships. Overall, iterative learning processes can achieve a much greater improvement in accuracy.

[0037] The method or system according to the invention acquires parameters or signals with which the accuracy of the output-side movement can subsequently be improved. The acquired signals or parameters can be applied directly to the system in which the determination was made, but can also be applied to technically similar systems by transferring the parameters or signals to another, technically identical or technically similar system. In the present description, a technically similar system is understood to mean a technical system in which the actual accuracy of the movement can be further improved by using the parameters or signals generated on another system.When transferring parameters or signals to another, technically identical or technically similar system, it is not absolutely necessary for the corrective parameters or signals for a movement to be generated by a self-learning controller. In such cases, the improvement in accuracy can also be achieved by using a corrective controller. This corrective controller intervenes in the system with the parameters or signals determined in advance and thereby improves the accuracy. However, the determination of the parameters or signals can be carried out by another, namely a self-learning controller. The corrective controller itself therefore does not need to have a learning function, but only needs to apply the correction. The signals or parameters generated by the self-learning controller can also be referred to as improvement factors or influencing factors.

[0038] Another characteristic of the method according to the invention is that it does not work for any desired value curves, but only for repetitive machine movements or runs, and that the correction values, in particular the generated parameters or signals, are calculated in advance for a complete machine movement, and that the correction does not take place immediately and in real time, but the correction values, in particular the parameters or signals, for each complete future machine movement or run are determined from the analysis of the errors from one or more past machine movements or runs.

[0039] In one embodiment, the method further comprises generating a synchronization signal temporally correlated with or based on a comparable target signal or comparable specification signal, and introducing the generated parameters or signals based on this synchronization signal.

[0040] In this way, it can be ensured that the correction signals are introduced at the right time, i.e. that the correction signals intervene at the right point in the repeated movement. At this point it should be noted that an exact match between the original "target signal" and the subsequent "comparable target signal" is not required in order to achieve an improvement in accuracy. The term "comparable target signal" should be understood to mean that using the correction signals still results in an improvement in accuracy. As long as the "comparable target signal" is sufficiently similar to the original "target signal", this improvement in accuracy occurs. This statement should be understood in both a spatial and temporal sense. Comparable target signals can therefore be used which have slight spatial changes compared to the previous target signals.However, comparable target signals can also be used which have slight temporal changes, for example a slightly slower movement sequence or a slightly accelerated movement sequence compared to the previous target signal.

[0041] By using the synchronization signal, the correction signal can be easily synchronized with the desired time course or with the desired position course, e.g. a path segment.

[0042] In one embodiment, the method further comprises iterative optimization of the generated parameters or signals by the self-learning controller by repeatedly applying the target signal and repeatedly incorporating the generated parameters or signals into the target signal or the comparable target signal and / or into the drive controller and / or into the drive signal for the electric drive. In particular, the above-mentioned acquisition, generation, incorporation, and movement execution steps can be performed multiple times.

[0043] This allows for efficient, high actual accuracy of the output-side movement. In particular, it allows for increasingly higher accuracy of the machine movement or the corresponding movement cycle to be achieved reliably.

[0044] In a further embodiment, the self-learning control is an iterative self-learning control, wherein the following steps are further carried out:

[0045] Execution of several output-side movements defined or described by the setpoint signal and by generated parameters or signals at the output of the transmission device in several successive runs,

[0046] Recording the output-side movements at the output of the transmission device during several or all runs with the measuring system, comparing the target signal, i.e. the specification or the specification signal, with the recorded output-side movements,

[0047] Adjusting the generated parameters or signals to increasingly improve the actual accuracy of the output-side movement after several or all of these runs based on at least one comparison of the target signal, i.e. the specification, with the detected output-side movement.

[0048] These measures can achieve a significant increase in accuracy, since a significant portion of the deviations is caused by the non-ideal properties of the drive trains, but these can be very well eliminated using iterative learning compensation methods.

[0049] At the same time, the use of secondary measuring systems on the mechanical transmission device offers an advantage in terms of mechanical installation compared to a separate external measuring system, for example a laser tracker mounted next to the machine or next to the robot, since the overall installation space for the entire system is smaller and the costs for external measuring systems can be saved.

[0050] In a further embodiment, the method further comprises transmitting the generated parameters or signals to a technically identical or similar system, in particular to a kinematically and dynamically comparable technical system, which technical system comprises a drive controller and / or a control device for a drive controller, an electric drive connected to the drive controller, and a mechanical transmission device connected to the electric drive, wherein the transmitted parameters or signals are used in the further, technically identical or similar system. It may also be expedient if the further, technically identical or similar system does not have a self-learning controller and / or a measuring system directly connected to the transmission device, in particular no secondary sensor outside or apart from the electric drive.

[0051] Transferring parameters and signals to other technically identical or similar systems offers significant advantages. While the achievable accuracy typically suffers when transferring to a similar system, it offers the advantage that the parameters only need to be determined once and then transferred to the desired target systems. Furthermore, with a technically similar system, it is no longer absolutely necessary to have a secondary measurement system. Although the similar system without a secondary measurement system can no longer "learn," the costs for the secondary measurement system can be saved.

[0052] In a further embodiment, the method comprises additionally controlling the system using a feedforward control based on a data-based model and / or additionally controlling the system using a feedback control based on sensor-detected deviations. According to a further development, the system control can be further supplemented by estimated state variables (observers).

[0053] The described method can also be combined with classic feedforward or feedback control structures. Classic feedforward compensation is based on a model, possibly supplemented by estimated state variables (observers), to compensate for influences on the output variable. Examples of such feedforward compensation include elastic deformations in the drivetrain, which can be modeled by the load torque at the output and an elastic characteristic of the drivetrain. This additional feedforward compensation is particularly advantageous during the learning phase, as it can shorten the system's learning time.

[0054] Feedback compensation calculates compensation signals based on sensor information—if necessary, supplemented by estimated state variables from a model. Feedback compensation can determine compensation for variable disturbances that were not present during the learning phase for the learning compensation or that are fundamentally variable—i.e., present differently during each run of the task. In this way, the influence of such disturbances can be reduced as much as possible. A particular advantage here is that long-term stability of the actual accuracy can also be ensured.

[0055] A combination of all methods allows for model-based deviations (feedforward compensation), data-based deviations (learning compensation), and sensor-based deviations (feedback compensation). Appropriate combinations of these compensation methods thus enable a further overall improvement in accuracy.

[0056] The object of the invention is further achieved by a system for generating machine movements. In particular, this system is intended for generating predetermined machine movements, preferably for generating repetitive machine movements that are to be repeated in several runs. In particular, the system is designed to carry out the method specified in the claims. This system comprises: a drive controller and / or a control device for a drive controller, wherein the control device can also be an integral component of the drive controller; an electric drive connected to the drive controller, in particular an electric motor, which drive controller can be controlled or supplied with a setpoint signal or specification signal, and which drive controller is designed to supply the electric drive with a drive signal;a mechanical transmission device connected to the electric drive, a measuring system, in particular a secondary sensor outside or away from the electric drive, which measuring system is directly or permanently connected to the transmission device, in particular is mechanically attached to the transmission device, and which measuring system is designed to detect an output-side movement at the output of the transmission device, a self-learning controller which is designed to generate parameters or signals to improve the actual accuracy of the output-side movement, which parameters or signals are generated based on the target signal and the movement detected by the measuring system, wherein the parameters or signals are preferably calculated in advance based on movement deviations determined by the self-learning controller,and which self-learning controller is further configured to incorporate the pre-generated parameters or signals (i) into the setpoint signal or into a setpoint signal of another system and / or (ii) into the drive controller or into a drive controller of another system and / or (iii) into the drive signal for the electric drive or into a drive signal for an electric drive of another system.

[0057] The effects and beneficial effects that can be achieved with this system can be found in the previous and following descriptions.

[0058] According to one embodiment, the system also comprises a synchronization signal generating device which generates a synchronization signal in a time-correlated manner with or based on a comparable target signal or comparable specification signal, wherein a corrective controller, in particular the self-learning controller, has a synchronization signal detecting device which controls the introduction of the generated parameters or signals based on the synchronization signal.

[0059] This ensures that the correction signals are applied at the correct time. By generating the synchronization signal (usually based on the preset signal) and using the synchronization signal (usually controlling the correction signal), the correction signal can be easily synchronized to the desired time profile or position profile, e.g., a path segment.

[0060] According to one embodiment, the measuring system is configured to detect at least one of the following measured variables: position, speed, acceleration, force, or their rotational equivalents, or a combination thereof. This allows for precise machine movement that corresponds as closely as possible to the specifications.

[0061] For smaller machines, particularly robots, drive components with such secondary encoders are available on the market. This makes it very easy to operate existing systems with increased accuracy. Larger machines, particularly robots, often have secondary encoders as an option. This makes it relatively easy to achieve improved accuracy by using the appropriate option. Another advantage is that when secondary encoders are used in some or all drive axes, all degrees of freedom of movement can be recorded and improved. In contrast, an external or peripheral laser tracker often only measures the position in Cartesian space and not the orientation, as this is significantly more complex. Likewise, an acceleration sensor only directly records a change in position and not a change in orientation.However, the restriction of the measured degrees of freedom can lead to portions of compensation signals being assigned to drive axes other than those corresponding to the cause in one drive axis. This can impair the convergence of the learning compensation. When using secondary encoders in all drive axes of a system, especially a machine or robot, the deviations occurring in each axis are immediately recorded and can thus be correctly assigned and compensated for immediately.

[0062] According to a further embodiment, the electric drive and the mechanical transmission device are configured in multiple units and are components of a multi-axis industrial robot or a multi-axis machine. This represents an advantageous application. According to one embodiment, the self-learning controller of the system can be configured to generate parameters or signals for improving the actual accuracy of the output-side movement exclusively for those electric drives and the associated mechanical transmission devices that are assigned in particular to the first three axes, i.e., the three axes closest to the robot base, of a multi-axis, in particular an at least six-axis, industrial robot. This makes it possible to achieve a favorable cost-benefit ratio.

[0063] In a multi-axis machine, the best accuracy improvement can be achieved if all drive systems or transmission devices are equipped with secondary measuring systems in combination with learning compensation. However, it is also possible, for example, to equip only those axes that contribute the most to deviations with this functionality, thus reducing the effort. Such an approach can be useful, for example, by equipping the first three axes of a 6-axis robot, which produce the largest proportion of path deviations, with the described functionality. In this way, a cost-optimized design can be created that meets the requirements of the application, depending on the application.

[0064] According to one embodiment, the self-learning controller is configured to intervene in the drive controller with its generated parameters or signals or to be functionally integrated therein and / or to intervene in a control device arranged upstream of the drive controller or to be functionally integrated therein, or to intervene in a signal path between the control device and the drive controller. According to the first alternative, efficient retrofitting of existing systems can be carried out, while also achieving stable functionality. According to the second alternative, functionally improved, technically particularly stable drive controllers and / or control devices can be created.

[0065] Another advantage of this type of use is that the compensation for the individual drives in a system can be determined independently of one another. The learning compensation can thus be carried out in an electronic controller that also calculates the setpoint curves for the individual motion axes. However, it is also possible to carry out the learning compensation in each of the drive controllers independently of one another. Furthermore, it would also be possible to set up hybrid systems in which parts are carried out in a central electronic controller and others in the drive controller. This is advantageous, for example, because the self-learning control, in particular the compensation control, can be implemented depending on the available computing capacity. If a central machine control system offers sufficient computing power, it can also be loaded with the compensation control.However, if the central machine control system is at its performance limit, the drive controllers could also take over this compensation control task. For simple axes, it is particularly advantageous if the compensation control is carried out in the drive controller, since an increase in accuracy can be implemented purely through an additional function of the drive controller.

[0066] In a further embodiment, the generated parameters can be stored in a remanent memory of the system, in particular in a remanent memory of the drive controller, or a control device for the drive controller, or of the motor, or of the transmission device, or of the measuring system. The advantage of storing the parameters in one or more of the specified memory locations is that the parameters are available again if the supply voltage fails. Furthermore, this makes it easy to service the system. If a component of the system is replaced, the required parameters are still available. The transmission device, e.g. a gearbox, is a particularly suitable storage location because this has the greatest influence on the parameters or signals.Since transmission devices typically do not have data storage, a secondary measuring system memory may be particularly suitable. To read the current status of the parameters or to make changes to them, it may also be useful to design the memory in such a way that, for example, read or write access to the parameters is possible via the controller. Depending on the application, access via the controller (e.g., the drive controller or the machine controller) or via a maintenance or diagnostic device is possible.

[0067] In a further embodiment, the self-learning controller is an iterative self-learning controller which is configured to execute the output-side movements at the output of the transmission device defined or described by the target signal and by generated parameters or signals in a plurality of successive runs, to record the output-side movements at the output of the transmission device during a plurality of or all runs with the measuring system, to compare the target signal, i.e. the specification or the specification signal, with the respectively recorded output-side movements, and to adapt the generated parameters or signals to increasingly improve the actual accuracy of the output-side movement after a plurality of or all of these runs based on the at least one comparison of the target signal, i.e. the specification, with the recorded output-side movement.

[0068] Based on previous tests, an ILC (Iterative Learning Control) is particularly well-suited as a self-learning controller in this type of application. It is characterized by high stability and excellent convergence behavior. This utilizes proven technology to achieve the highest levels of accuracy. However, the self-learning controller does not necessarily have to be designed as an ILC. It can also be based on a neural network, particularly a recurrent network. Fuzzy logic controllers, genetic algorithms, reinforcement learning, adaptive control, or similar controllers are also possible.

[0069] In a further embodiment, the system is operable in at least two modes, wherein in a first mode the self-learning control is active and in a further mode the self-learning control is inactive or introduces or provides other parameters and / or signals, in particular when different loads or movement sequences are provided.

[0070] Some applications have path sections or movement phases where accuracy is required to be very high, while other path sections have lower accuracy requirements. Furthermore, external conditions may change. For example, pick-and-place robots might move objects with varying degrees of inertia. It is therefore advantageous to design the high-precision control system so that it can be switched on, off, or toggled. This allows the application to be even better configured for an optimized process.

[0071] In a further embodiment, as briefly outlined above, the control device comprises a synchronization signal generating device that generates a synchronization signal that correlates in time with the target signal. Additionally, the self-learning controller comprises a synchronization signal detecting device that controls the introduction of generated parameters or signals based on the synchronization signal.

[0072] The synchronization signal generation device and its connection to the synchronization signal detection device make it easy to determine whether and, if so, which parameters or signals should be fed in for correction. The advantage here is that, in the simplest case, the correction can be switched on or off by applying a binary value. Furthermore, if the synchronization signal detection device is more complex, it is possible to switch between different correction modes. A time-correlated signal is particularly suitable for this purpose and can be designed to switch on at specific path segments, particularly when a correction is required. If a different path segment is traversed and a different correction is required, or if correction is no longer needed, the value can be changed accordingly or reset.The synchronization signal, which correlates in time with the target signal, is to be understood in such a way that the self-learning controller thereby has the option of applying the desired parameters or signals. This can occur both instantly and with a lead time. It is also not absolutely necessary for the synchronization signal generation device to generate the synchronization signal exclusively based on points in time. Rather, it is often more effective to provide a suitable synchronization signal based on the position of the target values ​​or actual values ​​of the path progress. However, these values, in turn, are temporally linked to the specified signal or target signal and thus correlate. The embodiment could also be applied analogously to a corrective controller without a self-learning function.

[0073] This design is particularly advantageous when the self-learning controller is integrated into a drive controller. This makes it easy to provide a drive controller whose accuracy can be easily increased externally if necessary.

[0074] In a further embodiment, the system further comprises an operating terminal which is preferably portable and mobile by a single operator, and which operating terminal is configured to generate the parameters and / or signals for improving the actual accuracy of the output-side movement, and / or is configured to adapt, store or delete generated signals and / or parameters, and / or is configured to transfer generated signals and / or parameters to technically identical or similar systems, and / or is configured to diagnose the system by incorporating the generated signals and / or parameters.

[0075] To configure a system for optimization, it is advantageous to provide the end customer or service technician with appropriate interaction options. These include determining parameters or signals, as well as transferring, deleting, editing, and / or diagnosing this data. The primary advantage is that the configuration of such a system is also accessible to the user or commissioning technician.

[0076] In a further embodiment, the operating terminal is further configured to visualize the data of at least one drive and the transmission device connected thereto in the following groups, for example in areas on a single display page of the operating terminal or individually on different display pages of the operating terminal:

[0077] - Learning data: Data of the learning process, e.g. movement path that is to be optimized,

[0078] - Measurement data: Data from the measurement system, e.g. data supplied by a secondary sensor, and

[0079] - Compensation data: Parameters or signals with which the improvement of the actual accuracy or compensation is carried out or can be carried out. Grouping can be done in such a way that the operator terminal provides a single display page with separate areas for the learning data, measurement data, and compensation data. Alternatively, the operator terminal can provide separate display pages for the learning data, measurement data, and compensation data.

[0080] By dividing the data into measurement data, learning data and compensation data, an intuitive and manageable structure is provided with which compensation data can be used or visualized independently of or detached from the learning procedure, and can also be generated externally.

[0081] According to a further embodiment, a technically identical or similar system, in particular a kinematically and dynamically identical or similar system, is provided. This system comprises a drive controller and / or a control device for a drive controller, an electric drive connected to the drive controller, in particular an electric motor, which drive controller can be controlled or supplied with a setpoint signal or specification signal. The drive controller is configured to supply the electric drive with a drive signal.The system further comprises a mechanical transmission device connected to the electric drive, as well as a corrective controller configured to process parameters or signals to improve the actual accuracy of the output-side movement, in particular to utilize parameters or signals obtained in advance using the method according to the claims. This corrective controller is further configured to incorporate the generated parameters or signals (i) into the target signal and / or into the drive controller and / or into the drive signal for the electric drive.

[0082] This embodiment does not necessarily include a measuring system or a self-learning controller. The measuring system could be omitted entirely, and a comparatively simpler corrective controller could be used instead of the previously used self-learning controller. However, parameters or signals for improving the actual accuracy of the output-side movement must first be obtained using a method according to the claims.

[0083] The advantage of this embodiment is that no measuring system is required for technically identical or similar systems. The self-learning controller can also be replaced by a more cost-effective corrective controller. Furthermore, it offers the possibility of easily transferring the correction signals of a reference system to numerous technically identical or similar systems with little effort. In particular, it may be expedient if the additional, technically identical or similar system does not have a self-learning controller and / or a measuring system directly connected to the transmission device, in particular, it does not have a secondary sensor outside or separate from the electric drive.

[0084] For a better understanding of the invention, it is explained in more detail using the following figures.

[0085] They show in a highly simplified, schematic representation:

[0086] Fig. 1A, 1B Systems known from the prior art for generating machine movements;

[0087] Fig. 2 shows an improved system for generating machine movements;

[0088] Fig. 3 shows a machine with several axes for the program-controlled execution of two- or three-dimensional positioning movements; Fig. 4 shows an operating terminal for use in the system for generating

[0089] Machine movements,

[0090] Fig. 5 the system according to Fig. 2 with additional synchronization measures.

[0091] By way of introduction, it should be noted that in the variously described embodiments, identical parts are provided with identical reference symbols or component designations. The disclosures contained throughout the description can be applied analogously to identical parts with identical reference symbols or component designations. Furthermore, the position information chosen in the description, such as top, bottom, side, etc., refers to the directly described and illustrated figure, and these position information must be applied analogously to the new position in the event of a change in position.

[0092] In the context of this description, the "machine axes" are to be understood as "controllable movement axes of the machine" or "positioning axes." These axes can be formed by joint axes, rotation axes, and / or linear axes.

[0093] Figures 1A and 1B illustrate known technical systems for generating machine movements in the form of block diagrams.

[0094] Each of these known systems comprises an electronic drive controller 1, which is configured to operate an electric drive 2, in particular an electric motor or a linear motor, connected to the drive controller 1. The drive controller 1 provides the electrical energy required for the scheduled operation of the electric drive 2.

[0095] Typically, the drive controller 1 is supplied with a preset or target signal 4 by a separate electronic control device 3, which at least partially describes the desired actuating movement of the system. Alternatively, or in combination with a separate control device 3, the control device 3 can also be a component integrated into the drive controller 1. The drive controller 1 is configured to supply the electric drive 2 with a drive signal 5.

[0096] The electric drive 2 is connected to a mechanical transmission device 6, which provides movement at the output when the drive 2 is activated. The mechanical transmission device 6 can comprise a gear 7, actuator arms, joints, linear guides, and other mechanical components, or can be a combination thereof. The transmission 6 can be a gear drive, a belt drive, a chain drive, or a combination thereof. The mechanical transmission device 6 provides movement at its output, i.e., on the downstream side, when the drive 2 is activated.

[0097] According to Fig. 1A, the system further comprises a first sensor 8, for example a position sensor, which is assigned to the drive 2 and detects the conditions prevailing at the output of the drive 2, for example an electric motor, such as the respective rotational angle positions. The corresponding detection values ​​of the first sensor 8, for example position values, are fed to the drive controller 1 via a signal path 9. Based on the respective detection values ​​and the target signal 4, the drive controller 1 can correct or compensate for any deviations at the output of the drive 2 within certain limits.

[0098] According to Fig. 1B, the system comprises, as an alternative or in combination with the first sensor 8 assigned to the output of the drive 2, a second or secondary sensor 10, which is assigned to the output of the mechanical transmission device 6 and is provided for detecting output-side states of the transmission device 6, for example positions. The corresponding detected values ​​of the secondary sensor 10, for example position values, are fed to the drive controller 1 via a signal path 11. Based on the respective detected values ​​and the setpoint signal 4, the drive controller 1 can correct or compensate for any deviations at the output of the transmission device 6 within certain limits. As already explained above, a combination with a first sensor 8 and a signal path 9 according to Fig. 1A is also possible, as indicated by dashed lines in Fig. 1B.

[0099] The accuracy of machine movements achievable with the known systems according to Fig. 1A, 1B is only partially satisfactory.

[0100] 2 and 3 illustrate an embodiment of an improved technical system with which the achievable movement or actual accuracy of at least one axis 12 of a controllable machine 13, in particular a multi-axis industrial robot, can be improved. For this purpose, the technical system comprises, among other things, an electronic, self-learning controller 14. In addition, an optimization or control method for the machine 1 is provided, as described below. This technical system for generating machine movements comprises a drive controller 1 and / or a control device 3 for a drive controller 1, an electric drive 2 connected to the drive controller 1, which drive controller 1 can be controlled with a target signal 4 and which drive controller 1 is configured to apply a drive signal 5 to the electric drive 2.Furthermore, the system comprises a mechanical transmission device 6 connected to the electric drive 2, for example a gearbox 7.

[0101] The system also includes a measuring system 15, in particular at least one secondary sensor 10 outside or away from the electric drive 2. The measuring system 15 or the at least one secondary sensor 10 is directly connected to the transmission device 6 or directly coupled thereto, in particular mechanically attached to the transmission device 6. The measuring system 15 is configured to detect an output-side movement at the output of the transmission device 6, for example, a two-dimensional (Cartesian) or spatial position or position change, using sensors or measurement technology.

[0102] The system also includes the aforementioned self-learning controller 14. This self-learning controller 14 is configured to generate parameters 16 or signals 17 to improve the actual accuracy of the output-side movement of the system. The parameters 16 or signals 17 generated by the self-learning controller 14 are further processed or used in the system using data or signal technology. In particular, the generated parameters 16 or signals 17 that improve the movement accuracy of the system are incorporated (i) into the setpoint signal 4 or into a comparable setpoint signal of another technically similar system and / or (ii) into the drive controller 1 or into a drive controller of another technically similar system and / or (iii) into the drive signal 5 for the electric drive 2 or into a drive signal for an electric drive of another technically similar system in a corrective or improving manner.

[0103] As indicated by dashed lines in Fig. 2, the accuracy-enhancing parameters 16 or signals 17 can be incorporated either into the setpoint signal 4, or into the drive controller 1, or into the drive signal 5 for the drive 2. However, a combination of these is also possible. To influence or modify the setpoint signal 4 or the comparable setpoint signal, the self-learning controller 14 can also be provided to act on the higher-level controller or the corresponding control device 3.

[0104] The self-learning controller 14 has been depicted as a separate unit, but it could also be implemented in the control device 3 and / or in the drive controller 1. In particular, the self-learning controller 14 can be configured to intervene in the drive controller 1 with its generated parameters 16 or signals 17 or to be functionally integrated therein and / or to intervene in a control device 3 arranged upstream of the drive controller 1 or to be functionally integrated therein, or to be configured to intervene in a signal path between the control device 3 and the drive controller 1.

[0105] The measuring system 15 on the transmission device 6, in particular its secondary sensor 10, for example in the form of a position or angle sensor, is connected to the self-learning controller 14 via the signal path 11, such that the measured values ​​of this secondary measuring system 15 can be transmitted to the self-learning controller 14 or made available to the self-learning controller 14. The self-learning controller 14 processes the measured data or measured variables acquired by the secondary measuring system 16 in combination with the at least one existing preset or target signal 4 and generates or calculates the compensation signals therefrom, in particular the accuracy-enhancing parameters 16 or signals 17.

[0106] The generated parameters 16 can be stored in a remanent memory 18 of the system, in particular in a remanent memory of the self-learning controller 14, of the drive controller 1, or of a control device 3 for the drive controller 1, or of the drive 2 or motor, or of the transmission device 6, or of the measuring system 15.

[0107] The measuring system 15 can be configured to measure at least one of the following measured variables: position, velocity, acceleration, force, or their rotational equivalents, or a combination thereof. These measured variables or measured data can be used to determine the respective actual position of the system or the respective axis 12 with respect to a two- or three-dimensional coordinate system.

[0108] It is expedient if the self-learning controller 14 is designed as an iterative self-learning controller (ILC). The ILC is configured to optimize the generated parameters 16 or signals 17 by repeatedly applying the setpoint signal 4 and repeatedly incorporating the generated parameters 16 or signals 17 into the setpoint signal 4 and / or into the drive controller 1 and / or into the drive signal 5 for the electric drive 2, thereby iteratively reducing or increasingly compensating for the system's movement deviations from the setpoint or target state.

[0109] In particular, the iterative self-learning control (ILC) of the system can be configured to execute the output-side movements defined by the target signal 4 and generated parameters 16 or signals 17 at the output of the transmission device 6 in several successive runs, i.e., to implement them multiple times. The output-side movements at the output of the transmission device 6 are recorded or monitored during several or all runs using the at least one secondary measuring system 15 on the at least one transmission device 6. The target signal 4 is compared with the respectively recorded output-side movements during all or individual runs, and any deviations are thus determined or calculated, which are used to generate or calculate the parameters 16 or signals 17.The generated parameters 16 or signals 17 are then adapted to increasingly improve the actual accuracy of the output-side movement after several or all of these runs based on the at least one comparison of the target signal 4 with the detected output-side movement in such a way that the movement deviations of the system are adaptively improved or gradually minimized.

[0110] The generated parameters 16 or signals 17, in particular the corresponding compensation data or compensation signals, can be used after the learning or generation phase in the technical system in which the learning or generation phase was carried out with the inclusion of the self-learning controller 14.

[0111] However, it is also possible to transmit the parameters 16 or signals 17 generated by a first system to another, technically identical or similar system (not shown). Such a technically identical or similar system also comprises a drive controller and / or a control device for a drive controller, an electric drive connected to the drive controller, and a mechanical transmission device connected to the electric drive. In such a technically identical or similar system, the transmitted parameters 16 or signals 17 are then processed or used to improve or increase the actual accuracy of an output-side movement of this system.However, in such a technically identical or similar system, it is not necessary to implement a self-learning controller, especially if the provided parameters or signals are used or processed by the control device or the drive controller. A technically identical or similar system exists if the parameters 16 or signals 17 generated by a first technical system lead to an improvement in the actual accuracy of the machine movements in the technically identical or similar system.

[0112] The machine 13 shown as an example in Fig. 3 comprises several axes 12 with which predetermined movements of at least one working point or of at least one end effector 19, for example, a welding torch, an application nozzle for paint or adhesive, a workpiece gripper, or the like, can be carried out automatically. The previously described system or method for improving the actual accuracy of machine movements can be used for all or individual axes 12 of such a machine 13.

[0113] The actuating or movement axes 12 of the machine 13 comprise, as is known per se, controllable drives 2, for example actuating or servo motors or linear drives such as piston-cylinder units. These drives 2 act on mechanical transmission devices 6, in particular on actuating mechanisms, such as articulated or telescopically mounted actuating arms. The at least one control device 20 of the machine 13 controls the corresponding drives 2 on the basis of a software-stored program sequence, wherein - as is known per se - a plurality of sensors or encoders can be integrated which influence the program sequence. An axis 12 or a combination of several axes 12, in particular of controllable drives 2 and mechanical transmission devices 6, can also be referred to as the drive train of the machine 13. The movement or actual accuracy achievable by the machine 13 with respect to a predetermined orplanned target movement path 21 - shown in Fig. 3 in full lines as an example as an S-curve between the spatial points PI and P2 - depends, among other things, on the elasticity of the actuating mechanisms of the axes 12, on the achievable positioning accuracies or positioning resolutions of the drives 2, on the precision of any brakes for the axes 12, on changing load moments or overhang widths 22, 23 of the end effector 19 relative to the base of the machine 13, on deviations in the transmission behavior of the drive train, on temperature influences, on movement speeds, on load changes, on mechanical vibrations, and on other factors.

[0114] As a result, actual deviations 24, which are considered unusable or undesirable beyond a certain extent—illustrated by dashed lines—may occur with respect to the two- or three-dimensionally predetermined target trajectory 21. These actual deviations 24 from the target trajectory 21, which occur during ongoing adjustment movements of the machine 3, can be reduced or compensated for with the optimization system or improvement method described above, in particular by means of the self-learning controller 14 and the secondary measuring system 15 on at least one of the mechanical transmission devices 6, within satisfactory, improved limit values.

[0115] In particular, it can be provided that the electric drive 2 and the mechanical transmission device 6 are configured in multiple units and are components of a multi-axis industrial robot. According to one embodiment, the self-learning controller 14 of this system can be configured to generate parameters 16 or signals 17 for improving the actual accuracy of the output-side movement exclusively for those electric drives 2 and the associated mechanical transmission devices 6, which are assigned in particular to the first three axes 12, i.e., the three axes 12 closest to the robot base, of a multi-axis, in particular an at least six-axis, industrial robot.

[0116] To further increase the accuracy of the machine movements, it can be provided that, in addition to the described accuracy improvement by a self-learning controller 14, the system is additionally controlled by means of a feedforward control based on a data-based model of the system. Alternatively, or in combination with this, it can be provided that the system is additionally controlled by means of a feedback control based on sensor-detected deviations, as illustrated in Fig. 1A or Fig. 1B. According to a further development, the control of the system can be further supplemented by state variables estimated by control technology, in particular by observers.

[0117] Fig. 4 illustrates an embodiment of an operator terminal 25 that can be used in the system or method for improving the actual accuracy of machine movements. The operator terminal 26 is preferably mobile or portable, so that it can be moved by an operator to different locations. However, a stationary embodiment is also conceivable. Stationary computers or mobile computers running a corresponding application for operating the system are also conceivable as the operator terminal 25.

[0118] The technical system can be operated in at least two modes, wherein in a first mode, the self-learning controller 14 - Figs. 2, 3 - is active, and in a further mode, the self-learning controller 14 is inactive or introduces or provides other parameters and / or signals, particularly when different loads or motion sequences are intended. For this purpose, at least one software- and / or hardware-implemented mode switch 26 can be provided.

[0119] The operating terminal 25 is configured to (i) generate the parameters 16 and / or signals 17 - Fig. 2 - to improve the actual accuracy of the output-side movement, (ii) and / or to adapt, store or delete generated signals 16 and / or parameters 17, (iii) and / or to transfer generated signals and / or parameters to technically identical or similar systems, (iv) and / or to diagnose the system by incorporating the generated signals and / or parameters.

[0120] According to an expedient embodiment, it can also be provided that the operating terminal 25 visualizes the data of at least one drive 2 and the transmission device 6 connected thereto - Fig. 2, 3 - in the following groups: Firstly, learning data 27, i.e. data of the learning sequence, such as the movement path that is to be optimized. Furthermore, measurement data 28, i.e. data of the measuring system 15, e.g. the data supplied by a secondary measuring sensor 10. Furthermore, compensation data 29, i.e. parameters 16 or signals 17 with which the improvement of the actual accuracy is carried out or can be carried out. This data can be grouped and visualized on a single display page of the operating terminal 25, in accordance with the embodiment according to Fig. 4. Alternatively, it is also possible to display this data individually on different display pages of the operating terminal 25.

[0121] The operator terminal 25 can comprise at least one safety switching element 30, for example, at least one emergency stop switching element 31 and / or at least one enabling button. This allows potentially safety-critical machine movements to be stopped immediately, or operator consent to execute potentially safety-critical machine movements can be given. Furthermore, the operator terminal 25 comprises at least one hardware and / or software-implemented operating element 32 for the user to influence or execute machine movements and technical systems.

[0122] Fig. 5 illustrates an embodiment of a system with synchronization devices, which can be used to additionally influence the application of correction signals. This system builds on the system illustrated in Fig. 2. The same reference numerals have been used for previously described components. A synchronization signal generating device 33 generates a synchronization signal 34 that correlates in time with the target signal 4 and matches the respective desired application (e.g., correction on / off). This synchronization signal 34 is detected by a synchronization signal detecting device 35 and, based thereon, controls the output of the desired parameters 16 or signals 17.

[0123] As already mentioned, temporal correlation is to be understood as the synchronization signal acquisition device 35 capturing the required signals in a timely manner (e.g., instantaneously or in advance) in order to be able to instruct the self-learning controller 14 accordingly. Depending on the desired application or modes, the self-learning controller 14 then supplies the appropriate parameters 16 or signals 17 and incorporates them into the system. The application could also be applied analogously to a corrective controller without a self-learning function.The embodiments show possible embodiments, whereby it should be noted at this point that the invention is not limited to the specifically illustrated embodiments thereof, but rather various combinations of the individual embodiments with each other are also possible and this possibility of variation lies within the skill of the person skilled in the art in this technical field due to the teaching of technical action by means of the objective invention.

[0124] The scope of protection is determined by the claims. However, the description and drawings must be used to interpret the claims. Individual features or combinations of features from the various embodiments shown and described may represent independent inventive solutions. The problem underlying these independent inventive solutions can be derived from the description.

[0125] For the sake of clarity, it should finally be pointed out that, in order to better understand the structure, some elements have been shown out of scale and / or enlarged and / or reduced in size.

[0126] Reference symbol list

[0127] Drive controller 32 Drive control element 33 Synchronization signal generation

[0128] Control device supply device target signal 34 synchronization signal

[0129] Drive signal 35 S y nchronization signal acquisition - transmission device sungseinrichtung

[0130] Gearbox first encoder

[0131] Signal path of secondary sensors

[0132] Signal path axis

[0133] Machine self-learning control

[0134] Measuring system parameters

[0135] Signals remanent memory end effector

[0136] Control device Target movement path Outreach width Outreach width Actual deviations Operating terminal Mode switch Learning data Measurement data Compensation data Safety switching element

[0137] Emergency stop switching element

Claims

P a t e n t a n s p r ü c h e 1. A method for improving the actual accuracy of an output-side movement of a system for generating predetermined machine movements that repeat in multiple passes, the system comprising an electric drive (2) with a mechanical transmission device (6) connected thereto, the method comprising the steps of: a) controlling a drive controller (1) of the electric drive (2) in a first pass with a setpoint signal (4), whereby a drive signal (5) is provided to the electric drive (2) and whereby an output-side movement is generated at the output of the transmission device (6), b) detecting the output-side movement with the aid of a measuring system (15) directly connected to the transmission device (6), in particular by means of a secondary sensor (10) outside or away from the electric drive (2), c) making the setpoint signal (4) available to a self-learning controller (14),d) making the detected movement available to the self-learning controller (14), e) generating parameters (16) or signals (17) to improve the actual accuracy of the output-side movement by the self-learning controller (14) based on the provided target signal (4) and based on the detected movement before the predetermined machine movement is repeated in at least one further run, f) incorporating these pre-generated parameters (16) or signals (17) by the self-learning controller (14) (i) into the target signal (4) or into a target signal of another system, and / or (ii) into the drive controller (1) or into a drive controller of another system, and / or (iii) into the drive signal (5) for the electric drive (2) or into a drive signal for an electric drive of another system,and g) executing at least one further run of the predetermined machine movement based on the drive signal (5) for the electric drive (2) modified in advance by the generated parameters (16) or signals (17).

2. The method according to claim 1, further comprising iteratively optimizing the generated parameters (16) or signals (17) by the self-learning controller (14) by repeated switching on of the setpoint signal (4) and repeated introduction of the generated parameters (16) or signals (17) into the setpoint signal (4) and / or into the drive controller (1) and / or into the drive signal for (5) the electric drive (2), in particular by repeated execution of the above-mentioned step b), optionally of step c), as well as steps d) to g) in direct temporal sequence of these steps and / or in non-direct temporal sequence of these steps.

3. Method according to claim 1 or 2, wherein the self-learning controller (14) is an iterative self-learning controller (ILC) and further the following steps are carried out: Executing several output-side movements defined by the target signal (4) and by generated parameters (16) or signals (17) at the output of the transmission device (6) in several successive runs, Recording the output-side movements at the output of the transmission device (6) during several or all runs with the measuring system (15), Comparing the target signal (4) with the recorded output-side movements, Adapting the generated parameters (16) or signals (17) to increasingly improve the actual accuracy of the output-side movement after several or all of these runs based on the at least one comparison of the target signal (4) with the detected output-side movement.

4. Method according to one of claims 1 to 3, further comprising transmitting the generated parameters (16) or signals (17) to another technically identical or similar system, in particular to a kinematically and dynamically comparable technical system, this further technical system comprising a drive controller (1) and / or a control device (3) for a drive controller (1), an electric drive (2) connected to the drive controller (1), a mechanical transmission device (6) connected to the electric drive (2), and Use of the transmitted parameters (16) or signals (17) in the technically identical or similar system.

5. The method according to claim 4, wherein the further, technically identical or similar system does not have a self-learning control and / or a measuring system directly connected to the transmission device, in particular no secondary sensor outside or away from the electric drive.

6. The method according to any one of claims 1 to 5, further comprising additionally controlling the system by means of a feedforward control based on a data-based model and / or additionally controlling the system by means of a feedback control based on sensor-based detected deviations.

7. The method according to claim 6, wherein the control of the system is further supplemented by estimated state variables.

8. System for generating predetermined machine movements, in particular for generating machine movements to be repeated in several runs, and in particular a system which is designed to carry out the method according to one of claims 1 to 7, the system comprising: a drive controller (1) and a control device (3) for the drive controller (1), an electric drive (2) connected to the drive controller (1), which drive controller (1) can be controlled with a setpoint signal (4) and which drive controller (1) is designed to apply a drive signal (5) to the electric drive (2), a mechanical transmission device (6) connected to the electric drive (2), a measuring system (15), in particular a secondary encoder (10) outside or away from the electric drive (2), which measuring system (15) is directly connected to the transmission device (6) and is designed toto detect an output-side movement at the output of the transmission device (6), a self-learning control (14) which is designed to provide parameters (16) or signals (17) to improve the actual accuracy of the output-side movement, generate which parameters (16) or signals (17) are generated on the basis of the target signal (4) and on the basis of the movement detected by the measuring system (15), wherein the parameters (16) or signals (17) are preferably calculated in advance on the basis of movement deviations determined by the self-learning controller (14), and which self-learning controller (14) is further configured to incorporate the parameters (16) or signals (17) generated in advance (i) into the target signal (4) or into a target signal of a further system and / or (ii) into the drive controller (1) or into a drive controller of a further system and / or (iii) into the drive signal (5) for the electric drive (2) or into a drive signal for an electric drive of a further system.

9. System according to claim 8, wherein the measuring system (15) is configured to detect at least one of the measured variables position, speed, acceleration, force or their rotational equivalents, or a combination thereof.

10. System according to claim 8 or 9, wherein the electric drive (2) and the mechanical transmission device (6) are designed multiple times and are components of a multi-axis industrial robot, in particular wherein the self-learning control of the system is set up to generate parameters (16) or signals (17) for improving the actual accuracy of the output-side movement exclusively for those electric drives (2) and the mechanical transmission devices (6) connected thereto, which are assigned in particular to the first three axes (12) of a multi-axis, in particular an at least six-axis industrial robot.

11. System according to one of claims 8 to 10, wherein the self-learning controller (14) is configured to intervene with its generated parameters (16) or signals (17) in the drive controller (1) or to be functionally integrated therein and / or to intervene in a control device (3) arranged upstream of the drive controller (1) or to be functionally integrated therein, or is configured to intervene in a signal path between the control device (3) and the drive controller (1).

12. System according to one of claims 8 to 11, wherein the generated parameters (16) can be stored in a remanent memory (18) of the system, in particular in a remanent memory of the self-learning controller (14), the drive controller (1), or a control device (3) for the drive controller (1), or the motor, or the transmission device (6), or the measuring system (15).

13. System according to one of claims 8 to 12, wherein the self-learning controller (14) is an iterative self-learning controller (ILC) which is designed to calculate the values ​​determined by the desired signal (4) and by generated parameters (16) or signals (17) to carry out defined output-side movements at the output of the transmission device (6) in several successive runs, to record the output-side movements at the output of the transmission device (6) during several or all runs with the measuring system (15), to compare the target signal (4) with the respectively recorded output-side movements, and to adapt the generated parameters (16) or signals (17) to increasingly improve the actual accuracy of the output-side movement after several or all of these runs based on the at least one comparison of the target signal (4) with the recorded output-side movement.

14. System according to one of claims 8 to 13, wherein the system is operable in at least two modes, wherein in a first mode the self-learning controller (14) is active and in a further mode the self-learning controller (14) is inactive or introduces or provides other parameters (16) and / or signals (17), in particular when different loads or movement sequences are provided.

15. System according to one of claims 8 to 14, wherein the control device (3) comprises a synchronization signal generating device (33) which generates a synchronization signal (34) which correlates in time with the desired signal (4), wherein the self-learning controller (14) comprises a synchronization signal detecting device (35) which controls the introduction of generated parameters (16) or signals (17) based on the synchronization signal (34).

16. System according to one of claims 8 to 15, further comprising an operating terminal (25) which is configured to generate the parameters (16) and / or signals (17) for improving the actual accuracy of the output-side movement, and / or is configured to adapt, store or delete generated parameters (16) and / or signals (17), and / or is configured to transfer generated parameters (16) and / or signals (17) to technically identical or similar systems, and / or is configured to diagnose the system by incorporating the generated parameters (16) and / or signals (17).

17. System according to claim 16, wherein the operating terminal (25) is further configured to visualize the data of at least one drive (2) and the transmission device (6) connected thereto in the following groups: - Learning data (27): Data of the learning process, e.g. movement path to be optimized, - Measurement data (28): data of the measuring system (15), e.g. data supplied by a secondary sensor (10), and - Compensation data (29): parameters (16) or signals (17) with which the improvement of the actual accuracy is carried out or can be carried out, in particular by grouping in such a way that the operating terminal (25) provides a single display page with separate areas for the learning data, measurement data and compensation data, or by the operating terminal (25) providing separate display pages for the learning data, measurement data and compensation data.

18. System for generating machine movements, the system comprising: a drive controller (1) and / or a control device (3) for a Drive controller (1), an electric drive (2) connected to the drive controller (1), which drive controller (1) can be controlled with a setpoint signal (4) and which drive controller (1) is designed to apply a drive signal (5) to the electric drive (2), a mechanical drive (2) connected to the electric drive (2) Transmission device (6), a corrective controller which is set up to process parameters (16) or signals (17) for improving the actual accuracy of the output-side movement, which parameters (16) or signals (17) were obtained in advance using the method according to one of claims 1 to 7, and which corrective controller is further set up to incorporate the generated parameters (16) or signals (17) (i) into the target signal (4) and / or (ii) into the drive controller (1) and / or (iii) into the drive signal (5) for the electric drive (2).

Images