Detecting machining errors of a laser machining system using deep convolutional neural networks

Patent metadata

Field	Value
Publication number	US-11536669-B2
Application number	US-201917295938-A
Country	US
Kind code	B2
Filing date	Oct 10, 2019
Priority date	Nov 22, 2018
Publication date	Dec 27, 2022
Grant date	Dec 27, 2022

Abstract

A system for detecting machining errors for a laser machining system for machining a workpiece includes: a detection unit for detecting image data and height data of a machined workpiece surface; and a computing unit. The computing unit is designed to generate an input tensor based on the detected image data and height data and to determine an output tensor on the basis of the input tensor using a transfer function. The output tensor contains information on a machining error.

First claim

The invention claimed is:

1. A system for recognizing a machining error for a laser machining system for machining a workpiece, comprising:
an imaging device for detecting raw image data of a machined workpiece surface;
a height sensor for detecting raw height data of a machined workpiece surface; and
computer;
wherein said computer is configured to create an input tensor based on the detected raw image data and raw height data, and to determine an output tensor based on the input tensor using a transfer function, the output tensor containing information about a machining error; and
wherein said input tensor comprises a two-channel image of the raw height data and raw image data,
wherein the transfer function between the input tensor and the output tensor is formed by a taught deep convolutional neural network,
wherein said taught deep convolutional neural network is adaptable to a changed situation using transfer learning in advance of a commissioning of the system.

2. The system according to claim 1 , wherein one or both of the imaging device and the height sensor comprises at least one of a camera system, a stereo camera system, an OCT system, and a triangulation system.

3. The system according to claim 1 , wherein the raw image data correspond to a two-dimensional image of a section of the machined workpiece surface.

4. The system according to claim 1 , wherein the raw height data correspond to a height geometry of the same section of the machined workpiece surface.

5. The system according to claim 1 , wherein the output tensor contains one of the following pieces of information: presence of at least one machining error, type of the machining error, position of the machining error on a surface of a machined workpiece, probability of a machining error of a certain type, and spatial and/or planar extent of the machining error on the surface of the machined workpiece.

6. The system according to claim 1 , wherein said computer includes an interface configured to receive training data for adapting said neural network and/or control data for determining the output tensor.

7. The system according to claim 6 , wherein the interface is configured to receive said training data and said training data comprise:
predetermined input tensors based on raw image data and raw height data of a machined workpiece surface detected by said imaging device and said height sensor; and
predetermined output tensors which are associated with the respective input tensors and contain information about existing machining errors of the machined work-piece surface.

8. The system according to claim 1 , wherein the input tensor has a dimension that is twice a number of the raw image data.

9. A laser machining system for machining a workpiece by means of a laser beam, said laser machining system comprising:
a laser machining head for radiating a laser beam onto a workpiece to be machined; and
a system according to claim 1 .

10. A method for recognizing a machining error in a laser machining system for machining a workpiece, said method comprising the steps of:
detecting raw image data and raw height data of a machined workpiece surface wherein raw image data is detected by an imaging device and raw height data is detected by a height sensor;
creating an input tensor based on the detected raw image data and raw height data, wherein said input tensor comprises a two-channel image of raw data of the height data and image data; and
determining an output tensor containing information about a machining error by means of a transfer function,
wherein the transfer function between the input tensor and the output tensor is formed by a taught deep convolutional neural network,
wherein said taught deep convolutional neural network is adapted to a changed situation using transfer learning in advance of a commissioning of the system performing the method.

11. The system according to claim 1 , wherein said input comprises a two-channel image consisting of an incident-light image and a height image.

Assignees

• Precitec Gmbh & Co Kg

Inventors

• Schwarz Joachim

Classifications

• G01N21/8851 Primary

  Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges (G01N21/8806 and G01N21/93 - G01N21/95692 take precedence; optical measurement of dimensions G01B11/00; optical scanning G02B26/10; image transformation G06T3/00; computerised image enhancement G06T5/00; image processing per se for flaw detection G06T7/0002) · CPC title

• B23K26/032 Primary

  using optical means · CPC title

• B23K26/21

  by welding · CPC title

• G01B9/02091

  Tomographic interferometers, e.g. based on optical coherence · CPC title

• G05B2219/33044

  Supervised learning with second artificial neural network · CPC title