DEVICE AND METHOD FOR DETERMINING THE STATUS OF A SPINDLE OF A MACHINE TOOL

Abstract

A device for determining a spindle status of a spindle of a machine tool includes a detector for detecting sensor data of the spindle for a defined time window. A processing unit analyses the sensor data through artificial intelligence by calculating a defined feature of the sensor data for the defined time window and determining a spindle status from the sensor data. An output member outputs the determined spindle status.

Claims

1.-15. (canceled)

16. A device for determining a spindle status of a spindle of a machine tool, said device comprising: a detector detecting sensor data of the spindle for a defined time window; a processing unit analyzing the sensor data through artificial intelligence by calculating a defined feature of the sensor data for the defined time window and determining a spindle status from the sensor data; and an output member outputting the determined spindle status.

17. The device of claim 16, wherein the determined spindle status characterizes a load of the spindle.

18. The device of claim 17, wherein the processing unit is configured to categorize the load of the spindle.

19. The device of claim 16, wherein the defined feature represents an average value of the sensor data or coefficients of a continuous wavelet transformation, or both.

20. The device of claim 16, wherein the defined feature represents an average value of the sensor data or the coefficients of a continuous wavelet transformation together with a statistical feature, or both.

21. The device of claim 20, wherein the statistical feature is a statistical torque or an absolute sum of changes within the defined time window, or both.

22. The device of claim 16, wherein the processing unit has stored therein an ensemble of at least ten decision trees comprising the sensor data, said processing unit being configured to ascertain the determined spindle status on the basis of the ensemble of at least ten decision trees.

23. The device of claim 16, the processing unit determines a data-driven load which approximates a real-time estimation and to which a spindle is exposed during operation.

24. The device of claim 16, wherein the processing unit stores an ensemble of at least ten decision trees comprising the sensor data, and wherein the spindle status is determined by the processing unit based on the ensemble of the at least ten decision trees.

25. The device of claim 16, wherein the processing unit analyses the sensor data by using a convolutional neural network.

26. The device of claim 16, further comprising a display, said output device being configured to transfer the determined spindle status to the display.

27. A machine tool, comprising: a spindle; and a device for determining a spindle status of the spindle, said device including a detector detecting sensor data of the spindle for a defined time window, a processing unit analyzing the sensor data through artificial intelligence by calculating a defined feature of the sensor data for the defined time window and determining a spindle status from the sensor data, and an output member outputting the determined spindle status.

28. The machine tool of claim 27, further comprising a display connected to the output member for visualizing the determined spindle status.

29. A method for determining a spindle status of a spindle of a machine tool, comprising: detecting with a detector sensor data of the spindle; analyzing the detected sensor data with a processing unit using artificial intelligence; determining a spindle status on the basis of the analysis of the detected sensor data; and outputting the determined spindle status.

30. The method of claim 29, further comprising categorizing by the processing unit a load of the spindle.

31. The method of claim 29, wherein the sensor data is detected for a defined time window.

32. The method of claim 29, further comprising: storing in processing unit an ensemble of at least ten decision trees comprising the sensor data; and ascertaining the determined spindle status with the processing unit on the basis of the ensemble of at least ten decision trees.

Description

[0045] The invention is described in more detail below on the basis of the exemplary embodiments described in the figures. In the drawings:

[0046] FIG. 1 shows an embodiment of a system comprising a spindle, a device and a visualization means, and

[0047] FIG. 2 shows a course of the method.

[0048] FIG. 1 shows an embodiment of a system comprising a spindle 1 of a machine tool 5, a device 2 and a visualization means 3. Sensor data of the spindle 1 is detected by the detecting means of the device 2 and analyzed in the processing facility of the device 2 by means of artificial intelligence. The processing facility determines the spindle status from the analysis. The output means of the device 2 transfers the determined spindle status to a visualization means 3. The visualization means 3 is configured as a display. The traffic light 4 on the visualization means 3 allows four categories to be displayed permanently allowed, allowed in the medium-term, allowed in the short-term and not allowed.

[0049] FIG. 2 shows a course of the method. Firstly signals Sges are detected by sensors. As examples of this, the temperature of the spindle S1, the torque of the spindle S2, the force of the spindle S(n1) and the speed of rotation of the spindle Sn can be detected. A larger number of signals is also possible, however.

[0050] A method step V1 relates to the time window already mentioned. Continuous sensor data is divided by means of time windows of a specific lengthpreferably between 0.5 seconds and 10 seconds.

[0051] In a method step V2, the detected sensor data is introduced as raw signals into a CNN, which is stored in the processing unit comprising artificial intelligence.

[0052] Alternatively, in a method step V3, a number of defined features for each signal, which is detected by a sensor, is calculated in the processing unit by means of artificial intelligence for a time window which is predetermined preferably at the factory or also by a customer and is subsequently no longer changeable and in a method step V4 these are transferred to an ensemble of regularized decision trees or boosted classification trees.

[0053] In a method step V5, the load level of the spindle is preferably output by means of the output means which is available in the inventive device.

[0054] In a method step V6, the load of the spindle is visualized, preferably categorized, by means of a display.

[0055] A method step V7 shows the already explained training of the artificial intelligence by means of recorded sensor data and a known load of the spindle.

[0056] The method indicates that a path which is based on a feature-based approximation and a path which is based on neural networks are implemented in the device respectively. The path via the CNN is mainly based on raw signals and can be extended or improved easily, by new data being inserted.