Machine tool control and method for feature map based error compensation on a machine tool

Abstract

The invention relates to a device and a method for compensating errors of a numerically controlled machine tool having at least one controllable machining axis for relative positioning of at least one workpiece with respect to one or more machining devices. The method comprises detecting actual measurement values of at least one input variable describing a state of the machine tool by means of sensors on the machine tool, providing at least one compensation parameter to be used by a control device of the machine tool to the control device of the machine tool, and compensating errors on the machine tool based on the compensation parameter provided to the control device. The compensation comprises compensation of thermal errors, compensation of geometric errors and compensation of errors based on mechanical dynamics.

Description

Technical Field

[0001] This invention relates to a method and apparatus for feature-map-based error compensation on machine tools. Furthermore, this invention relates to a system including a machine tool, particularly an NC or CNC machine tool, having machine tool control configured to compensate for errors based on feature-map-based compensation. Background Technology

[0002] Machine tool deformation and structure, as well as dynamic deviations, can have a decisive impact on machining accuracy. Various methods for compensating for deviations are known.

[0003] For example, DE 102014202878 A1 discloses a machine tool including a frame on which functional components that generate heat during operation are arranged, and a cavity structure is provided to form a circulation loop. A cooling medium circulates within the cavity of the frame to achieve temperature equilibrium between the warm and cold regions of the machine tool.

[0004] Furthermore, positional errors can be corrected with the aid of machine control. DE 102010003303 A1 relates to a method and apparatus for compensating for temperature-dependent positional variations on a machine tool having at least one linear axis. According to the invention, at least one temperature value is acquired at a temperature measurement location on the linear axis of the machine tool, a temperature difference between a reference temperature and the acquired temperature value is calculated, a compensation value is determined based on the temperature difference, and during the control of the machine tool, temperature-dependent positional variations (e.g., temperature-dependent displacements of a tool or workpiece clamped on the machine tool, a component of the machine tool, or the linear axis of the machine tool) are compensated based on the determined compensation value. Summary of the Invention

[0005] Regarding the aforementioned method and apparatus for correcting positional changes on machine tools, the object of this invention is to provide an improved apparatus and method for compensating for errors on machine tools, by means of which compensation can be performed more efficiently and accurately. Furthermore, the object of this invention is to provide improved machine control.

[0006] According to the present invention, as described in the independent claims, the above-described objectives of the invention are achieved. Features of preferred embodiments of the invention are described in the dependent claims.

[0007] According to one exemplary aspect, a method for compensating for errors in a CNC machine tool having at least one controllable machining axis for positioning at least one workpiece relative to one or more machining devices is proposed. The method may include: acquiring actual (or current measured) values ​​of at least one input variable describing the state of the machine tool (particularly spindle position, frame temperature, clamping path monitoring of a multi-clamp chuck, tool position, and spindle speed) using sensors on the machine tool; providing at least one compensation parameter at the machine tool's control unit to be evaluated by the control unit; and compensating for errors or deviations determined at the machine tool based on the compensation parameter provided to the control unit. The step of providing at least one compensation parameter may further include: providing one or more feature maps (particularly original feature maps), each of which can describe the structural characteristics of the machine tool and / or the geometric arrangement of multiple machine parts of the machine tool as a function of corresponding input variables; superimposing these maps to form a combined map; providing a global correction model configured to calculate the compensation parameter to be evaluated by the control device of the machine tool based on the provided combined feature map, acquired actual measurements and / or one or more control parameters of the control device, and calculating the compensation parameter using the global correction model based on the provided single feature map and / or combined feature map, acquired actual measurements and / or control parameters, and providing the compensation parameter to be evaluated by the control device at the control device of the machine tool. In this document, the compensation parameter can be the difference between the desired target state and the (measured or simulated) actual state. The compensation parameter can also indicate the difference between the current control parameters and the optimized control parameters, wherein the actual state of the machine tool can be adjusted / approximated to the target state using the optimized control parameters, and thus the machine tool error can be reduced or eliminated. Through the feature map-based compensation according to the invention, very fast and effective compensation for machine tool deviations is achieved. Therefore, it can significantly improve the machining accuracy of machine tools, while optimizing the computing power of machine control.

[0008] Feature-map-based compensation is advantageously extended to include volumetric compensation for errors (such as geometric errors (with respect to rotational and linear axes)) and compensation for errors caused by machine dynamics. Therefore, a comprehensive compensation based on one or more superimposed maps, directly at the machine control, is proposed. Thus, feature-map-based error correction in CNC machine tools is realized. Consequently, machining accuracy can be significantly improved without significantly increasing the computational and communication load on machine control.

[0009] Furthermore, error compensation can be performed directly at the machine control based on the combined feature map (i.e., error correction based on the feature map, such as positioning errors on the machine tool), wherein the combined feature map can be integrated into the global correction model of the machine tool. Particularly preferably, the combined feature map can be adjusted at least once or multiple times using a neural network. This has the following synergistic advantages: based on the provided feature map, compensation can be performed effectively, humanely, and extremely quickly and accurately in real time at the machine tool control. The feature map can be created based on finite element simulation combined with actual measurements (i.e., combining the simulation results of the machine tool with actual measurements to generate the corresponding feature map). Advantageously, the combined feature map can also be continuously optimized by the neural network. Therefore, the method for compensating for errors in CNC machine tools may further include the step of adjusting the combined feature map using a computer-implemented neural network (artificial neural network).

[0010] The global correction model preferably includes at least one feature map for vibration compensation of the machine tool. Compensation parameters for reducing vibration are determined by the global correction model based on the provided combined feature map (which includes the feature map for vibration compensation) and the acquired actual measurements. For vibration compensation, the error determined at the machine tool is compensated based on the compensation parameters provided to the control device. Particularly preferably, compensation parameters are determined for noise and vibration compensation (error compensation) of the machine tool. Herein, the compensation parameters are used to directly and actively influence the error at the machine tool by correspondingly controlling the machine tool via machine tool control. For this purpose, in addition to corresponding control of the machine tool's axes, active machine bearings and / or active absorbers are provided, for example, to reduce vibrations caused by machining during workpiece processing. Therefore, in CNC machine tools, errors can be compensated in an effective manner, thereby improving machining accuracy.

[0011] The provided feature maps can each describe the temperature characteristics, static displacement characteristics, and / or dynamic displacement characteristics of individual or multiple machine parts of a machine tool. Specifically, multidimensional feature maps are used, which are determined based on measurements and interpolation.

[0012] Combined feature maps can be provided by superimposing two or more feature maps. The feature maps can be selected from a set of maps, said set comprising three or more maps; for example, these maps may describe the corresponding temperature characteristics, static displacement characteristics, and / or dynamic displacement characteristics of single or multiple machine parts of a machine tool. Superimposing feature maps enables the provision of multidimensional feature maps that allow for a comprehensive description of the machine tool. Advantageously, feature maps used to compensate for geometric errors are superimposed with other feature maps (such as temperature feature maps, power feature maps, speed feature maps, etc.) to form combined feature maps.

[0013] Composite feature maps can be generated by superimposing or combining two or more feature maps, resulting in a feature map space spanning multiple dimensions. Feature maps can describe the temperature characteristics, static displacement characteristics, and dynamic displacement characteristics of a single machine part or multiple machine parts of a machine tool.

[0014] In addition to temperature values, input variables describing the state of the machine tool can also be position values ​​and / or acceleration values ​​and / or force values ​​and / or torque values ​​and / or strain values. Specifically, humidity measured on the workpiece, as well as dimensional and positional values, can also be included. It is further advantageous to use external measuring machines with traceability in measurement technology to further improve the accuracy of the feature maps.

[0015] The input variables for the feature maps and combined feature maps are preferably in the form of vector variables. Preferably, the multidimensional feature maps are first determined experimentally and then integrated into machine control. During operation, the multidimensional maps can be dynamically adjusted. Individual parameters of the feature maps can also be subsequently integrated into machine control to expand the combined feature maps. Therefore, the combined feature maps are configured to allow for subsequent expansion of the feature maps without significant effort. As the maps are expanded, new reference points can be added to the feature maps for feature map adjustments.

[0016] In a specific improvement of the invention, one reference point is preferably deleted before a new reference point is introduced, so as to always make full use of the memory space reserved for the feature map and to maintain the tunability of the feature map after long-term training. For this purpose, a reference point with minimal information content can be preferably selected. The absolute value of the error that occurs when a reference point is omitted and its position is replaced by interpolation is a measure of the information content of that reference point.

[0017] Compensation parameters can be calculated and / or error compensation based on the compensation parameters can be performed at predetermined time intervals. Feature maps can each be provided based on the acquisition of multiple reference measurements of the input variables and / or by evaluating a computer-implemented simulation model or by external measurements of the workpiece or tool.

[0018] Feature maps and / or combined feature maps can be adjusted based on acquired actual measurements and / or control parameters. Particularly preferably, the feature maps can be adjusted at least once using a neural network. This has the synergistic advantage that compensation can be performed efficiently, humanely, and very quickly and accurately in real time at the machine tool control center based on the provided feature maps. Particularly advantageously, feature maps can be created based on finite element simulation combined with actual measurements (i.e., combining the simulation results of the machine tool with actual measurements to generate the corresponding feature map). Advantageously, the combined feature maps can also be continuously optimized by a neural network. Feature maps and / or combined feature maps can be adjusted based on actual measurements and / or control parameters using a computer-implemented neural network or another AI algorithm. This particular combination enables the acquisition of comprehensive feature maps ideally suited for use with a global correction model, allowing for the output of a wide range of highly accurate correction values.

[0019] Adjusting the feature map may also include: reading at least a portion of the currently provided feature map and / or a portion of the currently provided combined feature map on a neural network or other AI algorithm; reading actual measurements and / or control parameters into the neural network or other AI algorithm; and determining the adjusted feature map and / or the adjusted combined feature map.

[0020] Adjustments to feature maps and / or combined feature maps can be performed automatically at predetermined time intervals.

[0021] At least one compensation parameter can be calculated using a global correction model based on adjusted feature maps or combined feature maps.

[0022] Actual measurement values ​​can be obtained when the workpiece is inserted into the machine tool.

[0023] Adjusting the combined feature map may further include: acquiring one or more geometric deviations between the target geometric dimensions of the first workpiece machined by the machine tool and the measured actual geometric dimensions, and adjusting the feature map and / or the combined feature map based on the acquired actual measurement values ​​and / or at least one control parameter and / or the acquired geometric deviations.

[0024] The actual geometry of the workpiece can be obtained through optical measuring devices and / or contact measuring devices.

[0025] The compensation parameters can be calculated using a global correction model by interpolating the combined feature maps based on the actual measured values ​​and / or control parameters obtained.

[0026] In addition, an evaluation device may be provided, which is configured to overlay feature maps to form a combined feature map and / or provide a global correction model, receive recorded actual measurements and / or control parameters, calculate compensation parameters based on the received actual measurements and / or control parameters, and provide the calculated compensation parameters to the machine tool's control device.

[0027] The evaluation device can be integrally formed as part of the control device. The feature map can preferably be recalculated directly at the control point. Alternatively, the recalculation can be performed in the cloud or on an external computer. The recalculated feature maps can be exchanged at the control point, or several feature maps can be stored and then selected accordingly.

[0028] A control system for compensating for errors in a machine tool may include: a control unit for the machine tool; one or more sensors attached to the machine tool to acquire actual measurements of one or more input variables describing the state of the machine tool; wherein the control system may be configured to perform a method for compensating for errors. The control system may be configured to correct errors on the machine tool in an integrated control manner and in real time. Particularly preferably, intelligent sensors that not only collect data but also interpret and transmit data are also provided as sensors. Furthermore, a machine tool including this control system is proposed.

[0029] The global correction model may also include a model based on physics that describes the elastic properties of the machine tool, including its structural variability. By using combined feature maps, it is also possible to efficiently determine the sensitivity of the correction to disturbances and the expected residuals. The method according to the invention also allows for continuous calculations during and between machine tool operations.

[0030] A method of operating a machine tool may include this compensation method, wherein drive signals can be corrected, and wherein at least one drive signal can be corrected by means of a characteristic curve or characteristic map, wherein the characteristic curve or characteristic map is determined for compensation depending on the operating point on a test bench or during machining testing. It is advantageous to use a characteristic map to correct at least one control signal. This allows for particularly effective controlled correction of the drive signals.

[0031] In several preferred exemplary embodiments, the neural network can control a feature map provided at the control unit of the machine tool. Sensor values ​​from one or more sensors of the machine tool (e.g., position measurements from a position measurement method performed on the machine tool) can be transmitted to the neural network as input data to adjust the feature map.

[0032] Measuring probes can be used as a position measurement method on machine tools. Furthermore, electromagnetic and / or optical measuring devices (e.g., laser measuring devices, camera devices) can be provided. Additionally, finite element models and simulations of the machine tool can be particularly preferably used to determine further position values.

[0033] Advantageously, the measuring device can be based on physical contact and / or configured as an optical measuring device and / or an ultrasonic measuring device and / or a radar measuring device and / or an RFID measuring device and / or a miniature GPS measuring device for detecting one-dimensional, two-dimensional and / or three-dimensional structures. The structure of the machine tool and the workpiece can be accurately obtained using the aforementioned measuring devices.

[0034] Further aspects and advantages, as well as the advantages and more specific implementation options of the foregoing aspects and features, will be apparent from the following description and explanation in connection with the accompanying drawings, which should not be construed as limiting in any way. Attached Figure Description

[0035] Figure 1 An exemplary diagram of a CNC machine tool 100 is shown;

[0036] Figure 2 The illustration shows the feature diagrams of various compensation methods;

[0037] Figure 3 The illustration shows feature map control optimized using feature maps. Detailed Implementation

[0038] Hereinafter, examples and exemplary embodiments of the invention are described in detail with reference to the accompanying drawings. The same or similar elements in the drawings will be denoted by the same reference numerals. It should be noted that the invention is not limited to the exemplary embodiments and implementation features described below, but also includes modifications to the exemplary embodiments, particularly within the scope of the independent claims.

[0039] According to exemplary embodiments of the present invention, compensation for geometric, static, dynamic, and thermal errors in a CNC machine tool 100 is proposed based on one or more superimposed feature maps. Therefore, it is proposed to adjust feature map-based compensation by volumetric compensation of errors (such as geometric errors (with respect to rotational and linear axes)) and compensation for errors caused by machine dynamics. Particularly advantageously, control of all corrections is proposed in a single correction model (the so-called global correction model). In this way, error correction in a precise feature map-based CNC machine tool is achieved with only reduced computational power. Therefore, machining accuracy can be significantly improved without significantly increasing the computational and communication effort of the entire machine control.

[0040] Figure 1The illustration shows a CNC machine tool 100 with machine tool control according to an exemplary embodiment of the present invention. By way of example, the machine tool is shown as a milling machine. However, the invention is not limited to milling machines, but can also be used with other machine tool types, such as metal cutting machine tools configured for machining workpieces by, for example, drilling, milling, turning, grinding, such as milling machines, general-purpose milling machines, lathes, turning centers, automatic lathes, milling / turning machine tools, machining centers, grinding machines, gear cutting machine tools, etc.

[0041] For example, machine tool 100 includes a frame that includes a machine bed 101 and a machine table 102. On the machine bed 101, for example, a movable machine skid 105 is arranged, which is mounted such that it can move horizontally in the Z direction (Z-axis) on the machine bed 101. For example, a workpiece WP is clamped on the machine skid 105, which may include, for example, a workpiece table. For this purpose, clamping devices may also be provided on the machine skid 105 or a tool table. Furthermore, in another exemplary embodiment, the machine skid 105 may include a rotary table that can rotate or pivot about a vertical and / or additional horizontal axis (optional axis of rotation or axis of rotation and / or pivoting). Additionally (or alternatively), the machine skid 105 can move in the horizontal Y direction (potentially perpendicular to the plane of the figures) by means of a Y-axis.

[0042] For example, a machine tool frame 102 carries a spindle carrier skid 103, which is vertically movable in the X direction on the machine tool frame 102, and a work spindle 104, for example, carrying a cutting tool, is held on the spindle carrier skid 103. The work spindle 104 is configured to drive a cutting tool WZ (e.g., a drilling and / or milling tool) held on the work spindle 104 to rotate about the spindle axis SA. For example, the spindle carrier skid 103 can be vertically moved in the X direction by means of the X-axis. Furthermore (or alternatively), the spindle carrier skid 103 can be moved in the horizontal Y direction (potentially perpendicular to the plane of the figures) by means of the Y-axis. Furthermore, in another exemplary embodiment, the spindle carrier skid 103 may include a rotation axis and / or a pivot axis to rotate or pivot the spindle 104 (optional rotation axis or rotation axis and / or pivot axis).

[0043] For example, machine tool 100 also includes machine tool control 200, which includes, for example, a screen 210 (e.g., configured as a touchscreen) and an input unit 220 for operation by the operator of machine tool 100. For example, input unit 220 may include means for user input or for receiving user commands from the operator, such as buttons, sliders, rotary controls, keyboards, switches, mice, trackballs, and possibly one or more touch-sensitive surfaces (e.g., touchscreens, which may be combined with screen 210). The operator can use machine tool control 200 to control the operation of the machine tool or the machining process on the machine tool, and also monitor the operating status of machine tool 100 or the machining process during machining. For example, machine tool control 200 includes NC control and programmable logic control (also SPS or PLC). In a preferred exemplary embodiment, the monitoring device may also provide data read or received by the machine tool control as input data to the neural network NN of the monitoring application, particularly periodically during workpiece machining. In a preferred exemplary embodiment, the monitoring device is configured to output control data affecting the machining process to the machine tool control system when the machine tool 100 is in an abnormal operating state, so as to interrupt the execution, stop the drive, or change the tool.

[0044] The errors to be compensated in the CNC machine tool 100 are not limited to thermal positioning errors, but include a large number of possible errors throughout the machine tool (i.e., the compensation according to the invention allows for the compensation of global errors in the machine tool). Specifically, these errors include positional deviations that dynamically cause movement of machine tool parts, such as spindle mounts, tool slides, machine tool slides 105, and main bearing carrier slides 103. Furthermore, the compensation according to the invention allows for active compensation for undesirable vibrations (and acoustic deviations) in the machine tool 100. Errors to be compensated in the CNC machine tool also include positional deviations caused by static factors of machine tool components, such as the weight of the clamped workpiece or varying weight distribution on the machine tool due to displaced machine tool components. According to the invention, a feature map-based control is proposed, preferably integrated into adaptive machine tool control. Compensation for geometric, static, dynamic, and thermal errors in the CNC machine tool 100 based on one or more superimposed feature maps is proposed.

[0045] exist Figure 2In this work, multiple feature maps are illustrated in a simplified form by way of example. Different feature maps are provided to compensate for different errors in a machine tool. For example, a corresponding feature map for machine tool 100 (or a machine tool component) can be provided for volumetric compensation of geometric errors (e.g., errors in linear and rotary axes). A corresponding feature map can be provided for dynamic compensation, to compensate for errors caused by the dynamics of the machine tool, such as pitch compensation. A corresponding feature map is also provided for compensating for temperature-related errors in the machine tool, such as errors caused by temperature increases related to machining or changes in ambient temperature. Advantageously, these feature maps can also be combined to form a combined feature domain. Thus, the feature map for compensating for geometric errors is advantageously superimposed with other feature maps (such as temperature feature maps, power feature maps, speed feature maps, etc.) to form a combined feature map. Therefore, independent and free correction of different error components of the machine tool can be achieved.

[0046] Characteristic maps can be represented, for example, as multiple characteristic curves or in a multidimensional coordinate system, where multiple characteristic curves are functions of multiple input variables. Characteristic curves describe the characteristics of a machine tool as a function of input variables. In practice, for example, characteristic curves are used to determine operating points to establish a linear approximation at specific points on the characteristic curve. Furthermore, they can be used to determine the power losses of components or to correct signals emitted by sensors.

[0047] For feature map-based compensation, the feature map can be understood as a continuous mapping of input variables to one-dimensional or multi-dimensional displacements or to compensation parameters. The most important input variables are parameters existing on the machine structure (e.g., recorded or measured by one or more sensors located at sensor positions on the machine tool structure, such as on the machine bed, on movable parts of the machine tool, on rotatable or pivotable parts of the machine tool, etc.) and position data of the machine tool axes. In addition to measurements, the feature map can be calculated or determined based on data processing on a virtual machine tool or simulation, or based on a digital twin of the machine tool and / or experimental measurements on the machine tool during test operations.

[0048] According to the present invention, an integrated application of a feature map (or a combination of multi-dimensional feature maps) is proposed, which is used to compensate for systematic geometric errors in linear or rotational axes, for compensating for temperature-related errors, and for compensating for errors caused by machine dynamics (such as pitch compensation). Correction terms are output from the combined feature map of the machine tool control system, which are predetermined using a large number of measurement points in the workspace and / or through computer simulations such as FEM (Finite Element Method) or multibody simulation. In other words, the feature map used to compensate for geometric errors is superimposed with other feature maps such as temperature feature maps, power feature maps, and speed feature maps to form a combined feature map. Therefore, independent and free correction of different error components of the machine tool can be achieved. Furthermore, the feature map can preferably be supplemented by creating parameters of mechanostatics (geometric accuracy) and / or dynamics (positioning accuracy). The feature map can advantageously be corrected / optimized by using a measuring device (measuring machine) to determine deviations at the part (initial sample) in order to precisely adjust the feature map to the range of the part.

[0049] For example, the combined feature map can be represented by multiple reference points existing in the machine tool's memory. Using overlay control, the deviation between the feature value read from the operating point and the optimal target value is determined. The combined feature map can be incorporated into a global correction model to determine compensation values. The global correction model according to the invention is configured to calculate compensation parameters to be evaluated by the machine tool's control unit based on the provided feature map, acquired actual measurements, and one or more control parameters of the control unit.

[0050] Based on a global correction model, a multidimensional correction vector, for example, for positional changes, can be output to correct errors (such as displacement correction). A significant advantage compared to conventional volume compensation is that compensation is performed using only a global correction model based on the provided feature map. This reduces computation time, enabling fast and efficient compensation. Furthermore, it makes control-independent and stable compensation possible.

[0051] In an advantageous further improvement according to an exemplary embodiment of the invention, based on the input variables, correction terms or one or more correction parameters for internal control compensation of errors can be periodically passed from the global correction model to the machine tool control, for example, by evaluating sensor and / or internal control data, and then the machine tool control performs error compensation using the transmitted correction terms. For example, geometric or dynamic errors can be compensated by adjusting the target positions of the linear axes, rotary axes, and / or pivot axes of the machine tool based on the transmitted correction terms. The errors of the CNC machine tool to be compensated are not limited to thermal positioning errors, but include a wide range of errors such as geometric errors, dynamic errors, and static errors of the machine tool (i.e., the compensation according to the invention allows for compensation of the global errors of the machine tool), etc.

[0052] To optimize the feature map (original feature map) and / or the global correction model, a neural network architecture that can update the feature map using new data is preferably implemented. The neural network can be provided externally by the machine tool control. However, it is particularly advantageous to integrate the neural network architecture into the actual machine tool control.

[0053] New training or input data for the neural network can preferably be determined from real processes or during operation on the machine tool (e.g., in experimental test operations and / or in actual machining processes on the machine tool), and additionally or alternatively, the training or input data for the neural network can also be determined from computer-executed simulations of the machine tool, such as on a virtual machine tool and / or on a machine tool-based digital twin. Together with existing data from finite element analysis, computer simulations of the machine tool (virtual machine tool or digital twin), and / or experimental operations of the machine tool, new data for optimizing feature maps can be generated with the help of the neural network. This optimization is preferably performed cyclically. Thus, a control integration solution for self-monitoring machine tools is realized.

[0054] For the adjustment of one or more feature maps by a neural network, the feature map or entries in the feature map can be passed from the machine tool control to the neural network as input data, and the feature map or at least a portion thereof can be read into the neural network. Furthermore, new data or input variables can be read from the machine tool, from sensors in the machine tool, and / or from the machine tool control. The feature map and / or at least one or more entries in the feature map are then updated or adjusted, preferably based on the network structure of the neural network. The network structure of the neural network can be based on radial basis functions and / or interpolation functions (e.g., linear and nonlinear regression methods). The continuous learning compensation algorithm according to the exemplary embodiment is preferably based on neural networks and / or interpolation functions (e.g., linear and / or nonlinear regression methods, radial basis functions, polynomial basis functions, etc.). Furthermore, a genetic algorithm for independently adapting interpolation reference points for the interpolation function can be incorporated. Therefore, independent and free correction of various error components of the machine tool can be achieved.

[0055] Compensation parameters are transferred from the feature map or global correction model of the machine control system. These parameters are predetermined by sufficient measurement points in the operating space and preferably by computer simulation. The feature map is also preferably updated or optimized over time to account for changes in the machine tool over time (e.g., wear) and thus compensate for changes in geometry over time.

[0056] Compensation parameters are used to adjust one or more target positions to correct or compensate for positional variations on a machine tool caused by geometry, dynamics, statics, and thermal factors. In this document, the corresponding compensation parameters can specify correction parameters applicable to individual axes of the machine tool, or correction parameters in individual orthogonal directions.

[0057] In a further exemplary embodiment, a real-time PLC system may be provided, for example, which reads input variables and outputs output variables. An IPC system connected to the PLC system may also be provided for index calculation. The IPC system processes the input data in index form, taking into account externally stored feature maps, so that compensation parameters are subsequently output to the machine tool via the real-time system based on the determined error or deviation.

[0058] In the first step, input variables are acquired, where the measured values ​​describe the state of the machine tool being detected and read. Multiple values ​​are read, such as temperature values ​​T1-T8, position value POS, etc., as input variables. Additionally, other sensor values ​​can be used as input variables, such as pressure measurements (e.g., pressure measurements in the machine tool's hydraulic and / or pneumatic systems, pressure measurements in the machine tool's cooling circuit system, etc.), oscillation or vibration measurements (e.g., from vibration or oscillation sensors), force measurements (e.g., from force sensors, strain gauge sensors, etc.), torque measurements (e.g., from torque sensors or calculations based on force measurements), active power values, acceleration measurements (e.g., from acceleration sensors), structural noise values ​​(e.g., from structural noise sensors), etc. Input variables are read at least once at the start of the process. Furthermore, it is advantageous to repeatedly perform reads at regular time intervals. Parameters obtained from or determined or evaluated by the machine tool control may include, for example, target position values ​​of the machine tool's axes, determined actual position values ​​of the machine tool's axes, rotational speed (e.g., spindle speed), motor current, motor power, etc.

[0059] The input variables read are subjected to rationality checks in the real-time system to ensure robust data processing and, for example, to detect measurement errors in an early stage. In a further step, compensation parameters are provided. Providing compensation parameters involves reading feature maps or combined feature maps and overlaying them to form a combined feature map. A global calibration model of the machine tool is created from the feature maps, and the compensation parameters are determined by this global calibration model. Specifically, the deviations of the input variables from the ideal parameters of the feature maps are recorded. Calibration or compensation parameters are calculated based on the recorded differences. The compensation parameters can be applied to the control values ​​of the machine tool via machine control so that errors in the machine tool are compensated, for example, by adjusting the target positions of one or more movable parts of the machine tool (e.g., linear axes, rotary axes, and / or pivot axes of the machine tool) based on the transmitted compensation parameters. For example, a position command can be compensated by adding a compensation parameter for the offset to the position command. Furthermore, the change between the last calculated offset magnitude and the current offset magnitude can be determined for each predetermined time.

[0060] According to some exemplary embodiments, a complete analysis of a machine tool in a data processing method or computer-based simulation begins with the analysis of one or more finite element models of the machine or machine tool, which can generate or calculate one or more feature maps preferably describing the temperature characteristics of the machine or machine tool. In further exemplary embodiments—as a supplement to or alternative to the calculation of one or more feature maps in an FEM-based simulation—the feature maps can also be calculated, determined, and / or adjusted based on a computer simulation of the machine tool (a so-called virtual machine tool or a so-called digital twin). Furthermore, or alternatively, the feature maps can also be determined and / or adjusted based on experimental measurements taken on the machine tool during test operations.

[0061] Figure 3 The paper demonstrates a feature map optimization method based on feature maps for error compensation on machine tools, using neural networks.

[0062] In the first step, parameters are determined or read out. Specifically, first parameters are determined and read out periodically at specific time intervals. These first parameters include, for example, temperature deviation, geometric deviation, and dynamic deviation. Additionally, second parameters are determined and read out. The second parameters are controlled and / or read out in real time. The second parameters include, for example, machine tool position, speed, power, rotational speed, etc. In the first step, input variables of the feature map are read out (e.g., read from machine tool control or by directly transmitting sensor values ​​from sensors on the machine tool).

[0063] Preferably, some or all of the input variables of the feature map to be updated are read. Additional input variables may also be read. Alternatively or additionally, sensor data may also be generated from a computer-implemented simulation of the machine tool (e.g., on a virtual machine tool or a machine tool-based digital twin), i.e., so-called synthetically generated sensor data. This preferably includes reading temperature measurements from the machine tool's temperature sensors. Furthermore, this may include reading position data related to the actual position of moving parts of the machine tool (e.g., movable axes), for example, from machine tool control or from the machine tool's position measurement sensors. Additionally, supplementary sensor data or control data may be read.

[0064] In the second step, one or more feature maps are used. Specifically, correction parameters are determined in real time based on the parameters determined in the first step. At least one feature map may be stored on an external computer or directly at the control center. The feature maps used include those describing the machine tool's structural characteristics (specifically, feature maps used to compensate for geometric errors), dynamic characteristics (speed feature maps), power feature maps, and temperature characteristics (temperature feature maps). Feature maps can be generated based on measurements and simulations based on finite element models, and are preferably provided as a stored data structure at the machine control unit (e.g., as a lookup table or lookup matrix). In addition to the aforementioned feature maps, specific workpiece feature maps of specific features can be advantageously read. Specific workpiece feature maps describe the machine characteristics when machining a particular predefined workpiece. Feature maps can also be combined to form combined (multidimensional) feature maps. This subsequently makes it possible to provide a global correction model, where all corrections can be mapped into a single model. On the control side, this results in higher accuracy with reduced computational power.

[0065] Advantageously, a specific temperature feature map, specifying temperature as a function of the measured input variable, is extended by a feature map virtually calculated through model computation. Furthermore, other measured and / or simulated feature maps, such as those used to compensate for geometric errors, power feature maps, and speed feature maps, can be combined. By incorporating feature maps (simulated and measured) into a combined feature map, the parameters contained in the virtual feature map can be adapted, enabling a global calibration model of the machine tool (which includes the interactions of all input variables) to achieve purely theoretical predictions and reports requiring minimal experimental support.

[0066] Advantageously, the second step may also include feature map adjustment, such as... Figure 3 As shown. First, training data for graph adjustment is determined. For example, training data can be determined through test procedures, machine measurements, or external workpiece measurements. The training data is then read to train a neural network or AI algorithm. Position data from external measurements can be used to compare the actual and target positions on the machine control unit with additional position measurements. Additionally, training data can be provided from computer-implemented simulations of the machine tool.

[0067] In the next step, the original feature map (the initial feature map before optimization) is loaded. The original feature map can be stored on an external computer or directly at the machine control unit. In the next step, at least one (original) feature map can be adjusted. Based on training data and previously determined parameters, an AI algorithm or neural network is used to adjust or optimize the map. In the final stage, the feature map (optimized feature map) can be provided to the machine control unit. For example, the optimized feature map is integrated into a global correction model and thus provided to the machine tool. The machine tool's control unit can use the global correction model, which includes the optimized feature map, to calculate the compensation parameters to be used.

[0068] In the final step, the correction value is transmitted to the machine control.

[0069] A global calibration model can be used to describe and correct the structural, dynamic, and thermal characteristics of the entire machine tool. First, the current state of the machine tool can be determined based on current measurements or input variables. For each state, there exists a separate set of parameters for a mapped feature map, where parameters are stored as functions of the input variable u. The parameters are determined based on the current input variable u(tn), for example, through interpolation. Errors on the machine tool can then be compensated based on compensation parameters provided to the control unit. For example, the global calibration model can be used to determine, for example, the displacement of TCP (tool center point; effective point of operation, tool reference tool operation point) at fixed or variable times. The calculated displacement can then be used as an offset. The global calibration model can be a control loop and preferably includes a black-box or white-box model of the machine tool. The global calibration model can also include links between various systems of the machine tool, whose output variables are measured by measuring elements and fed back to the machine control system via target / actual value comparison. The actuator, serving as the interface between the machine control and the controlled system, can be part of the controlled system.

[0070] In this paper, the global correction model is preferably directly integrated into the machine control, enabling autonomous compensation of deviations. Therefore, near-instantaneous error compensation is possible directly on the machine tool. By integrating compensation into the control, displacement can also be predicted based on the global correction model, where certain input variables, such as rotational speed, can be directly read from the NC code. Temperature characteristics, position, and displacement can also be predicted within certain constraints during machine tool operation, allowing the compensation parameters to be adapted to the expected error to minimize it. However, if further deviations are determined by evaluating measurements, the compensation parameters can be further adjusted. The global correction model can also be used to anticipate machine tool errors occurring during the measurement time, taking into account the time-varying characteristics of the machine tool. During the iteration process, coupling the measured / determined values ​​and value processing of input variables in the global correction model allows for optimal overall minimization of the residuals.

[0071] Further improvements involve generating and storing compensation parameters in association with the machine tool's system state, which is acquired during the generation of these parameters. When appropriate system conditions are later identified, the machine tool control can use the previously generated and stored compensation parameters to adjust the timing of machine tool function adjustments, thereby easily and effectively improving machine tool accuracy.

[0072] In a particularly advantageous further improvement, in addition to the aforementioned feature maps, specific workpiece feature maps are also used. These workpiece feature maps describe the machining characteristics when machining a specific, predefined workpiece. By integrating the workpiece feature maps into the combined feature maps, the accuracy of the machine tool when machining the defined workpiece can be significantly improved because specific workpiece deviations are already anticipated in the control. Therefore, these deviations or errors can be predetermined even before they actually occur, and appropriate compensation can be initiated by the machine control system even before the error occurs. Additional errors occurring during the machining process can be detected and compensated in real time.

[0073] Examples or exemplary embodiments of the invention and their advantages have been described in detail above with reference to the accompanying drawings. It should be noted again that the invention is by no means limited to or restricted to the exemplary embodiments and their implementation features described above, but also includes modifications to the exemplary embodiments, particularly those modifications that are included within the scope of the independent claims by modifying features of the described examples or by combining one or more features of the described examples.

Claims

1. A method for compensating for errors in a CNC machine tool (100), the CNC machine tool (100) including at least one controllable machining axis, the at least one controllable machining axis being used to position at least one workpiece relative to one or more machining devices, The method includes: The actual measured value of at least one input variable describing the state of the machine tool (100) is acquired by means of sensors on the machine tool (100). At least one compensation parameter is provided to the control device (200) of the machine tool, the at least one compensation parameter being evaluated by the control device (200) of the machine tool (100). The error on the machine tool (100) is compensated based on the compensation parameters provided to the control device (200). The step of providing at least one compensation parameter further includes: Multiple feature maps are provided, each of which describes the structural characteristics of the machine tool (100) and / or the geometric arrangement of the machine parts of the machine tool (100), as a function of the corresponding input variable. A global correction model is provided, which is configured to calculate the at least one compensation parameter to be evaluated by the control device (200) of the machine tool (100) based on the provided combined feature map, the acquired actual measurement values ​​and / or one or more control parameters of the control device (200). A combined feature map is provided by superimposing at least two feature maps from a plurality of provided feature maps, wherein each feature map in the superimposed at least two feature maps describes different characteristics of the machine tool (100) from a set of characteristics, wherein the set of characteristics includes: temperature characteristics, static displacement characteristics, dynamic displacement characteristics of individual machine parts of the machine tool (100), and dynamic displacement characteristics of a plurality of machine parts of the machine tool (100). The at least one compensation parameter is calculated using a global correction model based on the provided combined feature map, the acquired actual measurements, and / or control parameters. The control device (200) provides at least one compensation parameter, which is to be evaluated and calculated by the control device (200), to the machine tool (100).

2. The method according to claim 1, characterized in that, The provided feature diagrams each describe the temperature characteristics, static displacement characteristics, and / or dynamic displacement characteristics of one or more machine parts of the machine tool (100).

3. The method according to claim 1, characterized in that, The input variables describing the state of the machine tool (100) are at least temperature values ​​and / or additional position values ​​and / or acceleration values ​​and / or force values ​​and / or torque values ​​and / or strain values ​​and / or humidity values ​​and / or values ​​measured directly on the workpiece.

4. The method according to claim 1, characterized in that, The input variables of the feature map and / or the combined feature map are in the form of vector variables.

5. The method according to any one of claims 1-4, characterized in that, The calculation of the compensation parameters and / or error compensation based on the compensation parameters can be performed at freely selectable time intervals.

6. The method according to any one of claims 1-4, characterized in that, Each of the feature maps is provided based on multiple reference measurements of the input variables and / or by evaluating a simulation model implemented by a computer or by external measurements of the workpiece.

7. The method according to any one of claims 1-4, characterized in that, The feature map and / or the combined feature map are adjusted based on the actual measured values ​​and / or the control parameters obtained.

8. The method according to claim 7, characterized in that, Based on the actual measured values ​​and / or the control parameters, adjustments are made to the feature maps and / or the combined feature maps using a computer-implemented neural network or other AI algorithms.

9. The method according to claim 8, characterized in that, The adjustment of the feature map and / or the combined feature map includes: At least a portion of the currently provided feature maps and / or a portion of the currently provided combined feature maps are read into a neural network or other AI algorithm; The actual measured values ​​and / or control parameters are read into a neural network or other AI algorithm; Determine the adjusted feature map and / or the combined adjusted feature map.

10. The method according to claim 7, characterized in that, The feature map and / or the combined feature map are automatically adjusted at predetermined time intervals.

11. The method according to claim 7, characterized in that, The at least one compensation parameter is calculated using the global correction model based on the adjusted feature map or combined feature map.

12. The method according to any one of claims 1-4, characterized in that, The actual measured value is obtained when the workpiece is inserted into the machine tool (100).

13. The method according to claim 12, characterized in that, The adjustment of the feature map and / or the combined feature map also includes: Obtain one or more geometric deviations between the target geometric dimensions of the first workpiece machined by the machine tool and the measured actual geometric dimensions. The feature map and / or combined feature map are adjusted based on the actual measured values ​​and / or at least one control parameter and / or the acquired geometric deviation.

14. The method according to claim 13, characterized in that, The actual geometric dimensions of the first workpiece are obtained by means of an optical measuring device and / or a contact measuring device.

15. The method according to any one of claims 1-4, characterized in that, Based on the acquired actual measured values ​​and / or the control parameters, the calculation of the compensation parameters using the global correction model is performed by interpolation in the combined feature map.

16. The method according to any one of claims 1-4, characterized in that, The evaluation device is configured to overlay the feature maps to form the combined feature map and / or provide the global correction model, receive the acquired actual measurement values ​​and / or the control parameters, calculate the compensation parameters based on the received actual measurement values ​​and / or the control parameters, and provide the calculated compensation parameters to the control device of the machine tool (100).

17. The method according to claim 16, characterized in that, The evaluation device is integrally formed as part of the control device (200).

18. The method according to claim 15, characterized in that, The feature map is recalculated in the cloud or on an external computer and the recalculated feature map is exchanged at the machine control.

19. A control system for compensating for errors on a machine tool (100), comprising: Machine tool control device (200). One or more sensors, attached to the machine tool, for acquiring actual measurements of one or more input variables describing the state of the machine tool (100). The control system is characterized in that it is configured to perform the method according to any one of claims 1-18.

Images