DE102021207130A1 - System, method and computer program for the automated assessment of at least one component - Google Patents

Abstract

System (10) for the automated assessment of at least one component (B) comprising an imaging device (1), designed to image at least one region of the component (B); a guiding device (2) adapted to position the component (B) relative to the imaging device (1); an evaluation device (3) designed to run a computer program that reads in an image of the imaging device (1), classifies and/or localizes defects in the component (B) and outputs the defects when the computer program runs on the evaluation unit (3).

Description

The invention relates to a system, method and computer program for the automated diagnosis of at least one component

the EP 3 757 869 A1 discloses a method for determining and displaying potential damage to objects on overhead lines. According to one aspect, infrastructure elements of the overhead line such as masts and components such as fittings and add-on elements are determined from a point cloud data set determined by means of a laser scanning device about the overhead line and its surroundings, which also includes the exact position of the individual points, and representations of these elements for recognizable damage such as fractures , splintering but also foreign bodies such as ice coverings or vegetation are analysed. Machine learning methods known from the prior art can be used for this purpose.

Furthermore, automated and high-precision product troubleshooting methods based on artificial intelligence for non-destructive testing methods are known, see for example https://weisswaszu.de/Digitalisierung/kuenstliche-intelligenz-sucht- Fehler-inteile/.

the EP 2 609 460 B1 discloses a method for automatically focusing biological material in fluorescence microscopy.

Visual assessment is also a standard when assessing components, for example after tests have been carried out. In the known state of the art, a high level of manual effort is required for recurring assessments when assessing the component. The comparability of different findings can be impaired, for example, by different lighting conditions.

Proceeding from this, the object of the invention was to provide a standardized and automated assessment of components in development or pre-series.

The subjects of the independent and subordinate claims each solve this problem.

According to one aspect, the invention provides a system for the automated assessment of at least one component. The system includes an imaging device configured to image at least a portion of the component. Furthermore, the system includes a guide device, designed to position the component relative to the imaging device. The system also includes an evaluation device, designed to run a computer program that reads in an image of the imaging device, classifies and/or localizes defects in the component and outputs the defects when the computer program runs on the evaluation unit.

According to a further aspect, the invention provides a method for the automated assessment of at least one component. The component is automatically positioned relative to an imaging device. The imaging device images at least one area of the component. In the image obtained, defects in the component are classified and/or localized by executing a trained machine learning program.

According to a further aspect, the invention provides a computer program for the automated assessment of at least one component. The computer program includes instructions that cause a computer to execute a method according to the invention when the computer executes the computer program.

Advantageous refinements of the invention result from the definitions, the dependent claims, the drawings and the description of preferred exemplary embodiments.

The instructions of the computer program according to the invention include machine instructions, source text or object code written in assembly language, an object-oriented programming language, for example C++, or in a procedural programming language, for example C. The computer program is entirely or partially on an integrated circuit or in a remote system comprising a cloud executed. According to one aspect, for example an FPGA chip is configured using hardware program instructions to carry out the method steps.

The invention also relates to a data carrier on which the computer program according to the invention is stored. The data carrier includes volatile memory, for example RAM, DRAM, SRAM, and non-volatile memory, for example ROM, flash EEPROM. The data carriers are, for example, flash memory cards, USB sticks.

The invention also relates to a data carrier signal which transmits the computer program according to the invention. The data carrier signal is transmitted, for example, using radio technology including 5G technology or optically. This allows software to be provided over-the-air.

With the system according to the invention components in development, for example after tests carried out can be assessed automatically, i.e. without manual effort. In comparison to the known state of the art, no great effort is required for recurring findings.

Components include machine parts, for example parts of drive trains of a vehicle, for example transmission parts, and electronic parts, for example parts of hardware modules, for example control units.

According to one aspect, the diagnosis is an optical diagnosis. The imaging device includes, for example, a camera system comprising optics and an imaging sensor. The imaging sensor measures, for example, electromagnetic radiation reflected by the component, including electromagnetic radiation in the visible and/or infrared range. According to a further aspect, the imaging sensor is a time-of-flight sensor, a radar sensor or a lidar sensor. In this way, depth information of the component is obtained. According to a further aspect, the imaging device includes a combination of camera, radar and/or lidar sensors. According to a further aspect, a radiation device is integrated into the imaging device. The radiation device irradiates the component. This allows the reflection properties of the component to be evaluated.

For example, the guide device moves the component in one or more dimensions, for example in two dimensions within one plane. According to a further aspect, the guide device is designed to reproducibly position the component relative to the imaging device over a large displacement path in the millimeter to micrometer range.

According to one aspect, the evaluation device is embedded in the system. For example, the evaluation device is a microcontroller of the system. According to a further aspect, the evaluation device is a remote device. According to a further aspect, the evaluation device includes a memory into which the classified and/or localized defects in the component are written for documentation purposes. According to a further aspect, the evaluation device comprises an interface, for example a radio or wired interface, to such a memory. Defects include problem areas.

According to a further aspect, the evaluation device is designed to classify and/or localize the defects based on a target/actual comparison of images of the component. Defects can be clearly classified and/or localized by means of the target/actual comparison. According to one aspect, the evaluation device is designed to look up target images of the component and to compare them with the actual images. The target images are stored, for example, in an internal memory of the evaluation device or are loaded from a cloud memory. According to one aspect, the target/actual comparison is carried out pixel-wise or bit-wise.

According to a further aspect, the evaluation device is designed to classify and/or localize the defects by executing a trained machine learning program. According to one aspect, the trained machine learning program has learned target images in a data-driven manner and automatically carries out the target/actual comparison without accessing stored target images. For example, the machine learning program was trained with a number of target images of different components under different environmental conditions, for example under different lighting conditions. The machine learning program includes a support vector machine, a random forest classifier or an artificial neural network. In one aspect, the machine learner is a convolutional network.

According to a further aspect, the system comprises a radiation device for irradiating the component. The system is designed to position the radiation device relative to the component and/or relative to the imaging device. The result of this is that a component is repeatedly aligned in the same position relative to the imaging device under the same radiation conditions. This makes the findings reproducible.

The radiation device is, for example, a light source, for example an LED or an array of LEDs. For example, different findings are not affected by different lighting conditions.

According to a further aspect, the system comprises at least one control device for automatically controlling the imaging device, the guiding device and/or the radiation device.

According to one aspect, the control device comprises axle drives, for example motor-driven ball screws, other screw drives or also linear motors.

According to one aspect, the control device is designed to regulate and/or based on the images of the imaging device and/or based on signals from the evaluation device to determine control signals for the guiding device and/or the radiation device.

According to a further aspect, the guide device comprises at least one compound table or one robot arm.

A cross table, also known as an XY table or cross support, is a two-axis system consisting of two single-axis linear guide systems that enable an object to be moved in two directions within one plane. The two guides of the cross table determine the directions of movement in the X and Y directions. They are incorporated crossed. The angular offset is generally essentially 90 degrees. According to one aspect, the compound table is a tilting table in which this angle is adjustable. Thanks to the interaction of the two axes, the component can assume any position in the area of the X-Y plane, as long as the position is within the range of possible movement of the guides, i.e. the travel path.

According to one aspect, the robot arm is an articulated arm robot with several degrees of freedom, for example 4, 5 or 6. According to one aspect, the robot arm is designed to carry loads of up to 10 kg.

According to one embodiment of the method, the component is automatically positioned relative to a radiation device that illuminates the component. This increases the degree of automation.

According to a further aspect, a system according to the invention is used to carry out a method according to the invention.

According to one aspect, the defects in the component classified and/or localized in the method are stored for documentation purposes.

• 1 ein schematische Darstellung eines erfindungsgemäßes System und
• 2 eine schematische Darstellung eines erfindungsgemäßen Verfahrens.

The invention is illustrated in the following exemplary embodiments. Show it:

• 1 a schematic representation of a system according to the invention and
• 2 a schematic representation of a method according to the invention.

In the figures, the same reference symbols denote the same or functionally similar reference parts. For the sake of clarity, only the relevant reference parts are highlighted in the individual figures.

This in 1 The illustrated exemplary embodiment of a system 10 according to the invention comprises a camera as the imaging device 1. The imaging device 1 makes an image of a component B. Based on the image, an optical finding of the component B is obtained.

Furthermore, the system 10 includes a cross table as a guide device 2. The guide device 2 automatically positions the component B in a detection range of the imaging device 1 at a specific point and thus allows recurring and reproducible optical findings. In an alternative embodiment, the guide device 2 is an articulated arm robot.

The system 10 also includes an evaluation device 3. The evaluation device 3 reads in the image of the component B obtained with the imaging device 1, for example via a data interface to the imaging device 1. The evaluation device 3 processes the image using an artificial neural network that is trained to classify defects in components B and/or to localize them on component B. The trained artificial neural network outputs the classified and/or localized defects, for example via an output interface to a data memory for documentation purposes or to a display device.

The system 10 also includes a radiation device 4, for example a light source. The radiation device 4 irradiates the component B under the same radiation conditions in order to ensure, for example, the same lighting conditions when imaging the component B with the imaging device 1 .

The system 10 also includes a control device 5. The control device 5 controls the imaging device 1, the guiding device 2 and/or the radiation device 5 in order to automatically position the component relative to the imaging device 1 and/or the radiation device 5.

In 2 the component B is automatically positioned relative to the imaging device 1 in a method step V1. In a method step V2, the imaging device 1 images at least one area of the component B. In a method step V3, defects in the component B are classified and/or localized in the image obtained by executing a trained machine learning program.

reference list

B

component

10

system

1

imaging facility

2

guide device

3

evaluation device

4

radiation device

5

control device

V1-V3

process steps

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• EP 3757869 A1 [0002]
• EP 2609460 B1 [0004]

Claims (10)

System (10) for the automated diagnosis of at least one component (B). • an imaging device (1) designed to image at least one region of the component (B); • a guiding device (2) designed to position the component (B) relative to the imaging device (1); • an evaluation device (3), designed to run a computer program that reads in an image of the imaging device (1), classifies and/or localizes defects in the component (B) and outputs the defects when the computer program runs on the evaluation unit (3). system (10) according to claim 1 , wherein the evaluation device (3) is designed to classify and/or localize the defects based on a target/actual comparison of images of the component (B). System (10) according to one of the preceding claims, wherein the evaluation device (3) is designed to classify and/or localize the defects by executing a trained machine learning program. System (10) according to any one of the preceding claims comprising a radiation device (4) for irradiating the component (B), wherein the system (10) is designed, the radiation device (4) relative to the component (B) and / or relative to the To position imaging device (1). System (10) according to one of the preceding claims, comprising at least one control device (5) for automatically controlling the imaging device (1), the guiding device (2) and/or the radiation device (4). System according to one of the preceding claims, in which the guide device (2) comprises at least one cross table or one robot arm. Method for the automated assessment of at least one component (B), the component (B) being automatically positioned (V1) relative to an imaging device (1), the imaging device (1) imaging at least one area of the component (B) (V2) and defects of the component (B) are classified and/or localized (V3) in the image obtained by executing a trained machine learning program. procedure after claim 7 , wherein the component (B) is automatically positioned relative to a radiation device (4) that illuminates the component (B). procedure after claim 7 or 8th , wherein for carrying out the method, a system (10) according to one of Claims 1 until 6 is used. Computer program for the automated diagnosis of at least one component (B) comprising commands that cause a computer to carry out a method according to one of Claims 6 until 8th executes when the computer executes the computer program.

Images