WO2021219515A1 - Procédé, dispositif et programme informatique de génération d'informations de qualité concernant un profil de revêtement, procédé, dispositif et programme informatique de génération de base de données, et dispositif de surveillance - Google Patents

Procédé, dispositif et programme informatique de génération d'informations de qualité concernant un profil de revêtement, procédé, dispositif et programme informatique de génération de base de données, et dispositif de surveillance Download PDF

Info

Publication number
WO2021219515A1
WO2021219515A1 PCT/EP2021/060712 EP2021060712W WO2021219515A1 WO 2021219515 A1 WO2021219515 A1 WO 2021219515A1 EP 2021060712 W EP2021060712 W EP 2021060712W WO 2021219515 A1 WO2021219515 A1 WO 2021219515A1
Authority
WO
WIPO (PCT)
Prior art keywords
coating
quality information
image data
coating profile
database
Prior art date
Application number
PCT/EP2021/060712
Other languages
German (de)
English (en)
Inventor
Christian STOLZE
Jens Hagemann
Kai Tacke
Lars FÖLSTER
Original Assignee
Volkswagen Aktiengesellschaft
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Volkswagen Aktiengesellschaft filed Critical Volkswagen Aktiengesellschaft
Publication of WO2021219515A1 publication Critical patent/WO2021219515A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • G06F18/2178—Validation; Performance evaluation; Active pattern learning techniques based on feedback of a supervisor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/40—Software arrangements specially adapted for pattern recognition, e.g. user interfaces or toolboxes therefor
    • G06F18/41—Interactive pattern learning with a human teacher
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/98—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
    • G06V10/993—Evaluation of the quality of the acquired pattern
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection

Definitions

  • Method, device and computer program for generating quality information about a coating profile Method, device and computer program for generating a database, monitoring device
  • the present invention relates to a method, a computer program and a device for generating quality information, to a method, a computer program and a device for generating a database and a monitoring device, in particular, but not exclusively, to a concept for the automated quality control of coating profiles.
  • micrographs of APS layers are assessed manually by trained experts, which represents a high level of personnel expenditure, especially if checks are carried out regularly and frequently.
  • test results are very person-dependent and therefore not always comparable between different testers.
  • the document WO2019099928 describes an automated manufacturing system and manufacturing method for thermal and mechanical components using hybrid direct laser sintering, direct metal laser sintering, CNC (from “Computerized Numerical Control”), thermal spraying, direct metal deposition and friction stir welding.
  • the document DE 102018 128478 A1 discloses a method for coating a component using a robot spray system.
  • the robotic spray system includes a scanning device operable to measure and store properties of a surface before and after coating; a robotic arm operable to move the robotic spray system relative to the surface of the component, the component including one or more reference features that remain uncoated during coating; a spray nozzle operable to apply a spray coating to the To apply surface; and a device driver module including circuitry configured to operate the scanning device, the robotic arm, and the spray nozzle.
  • Embodiments are based on the core idea that in the quality control of coating profiles, artificial intelligence can be used in the assessment of coating profiles.
  • a database can be used on the basis of which an artificial intelligence, for example a neural network, can be trained and then used to assess coating profiles.
  • Embodiments create a method for generating quality information about a coating profile.
  • the method comprises a sensory acquisition of image data about the coating profile and an evaluation of the image data using artificial intelligence in order to obtain the quality information about the coating profile.
  • Exemplary embodiments can at least partially automate an evaluation process of image data through the use of artificial intelligence and thus make it more efficient.
  • the artificial intelligence can be trained based on a database with quality information about coating profiles. Training with a database can have the effect that at least all of the coating profiles contained in the database are correctly assessed by the artificial intelligence.
  • a quality of the assessment for example in the form of an error rate, can thus be influenced by the database.
  • the method can include training the artificial intelligence based on an assessment of the quality information by a user.
  • the artificial intelligence can be further trained or improved by such feedback from a user.
  • the sensory detection can include detection by means of a laser, an X-ray device and / or an optical detection method. This makes it possible to automatically assess image data of different origins.
  • the evaluation of the image data can include an assessment of at least one element from the group of a layer thickness, a layer transition, a defect, a pore, a void, a course of a layer transition, an oxidation and a foreign material.
  • a wide variety of errors or irregularities in the coating profile can be detected in an automated manner.
  • the quality information about the coating profile can be stored in a database.
  • the database can become more extensive over time.
  • the image data can be saved via the coating profile in the database. This enables subsequent control or assessment by appropriate experts. As the size and / or quality of the database grows, the training quality and thus the assessment quality of the artificial intelligence can be increased.
  • the coating profile can be, for example, a coating profile of a raceway in a cylinder crankcase, and the image data can include a micrograph of the coating profile.
  • Embodiments can thus create an automated assessment of coating profiles of cylinder crankcases.
  • Embodiments also create a method for generating a database with quality information about coating profiles.
  • the method comprises sensory acquisition of image data about the coating profile and evaluation of the image data by a user in order to obtain the quality information about the coating profile.
  • the method includes storing the quality information about the coating profiles in the database.
  • the database can thus be a well-founded training basis for the artificial intelligence described above.
  • the method can also include receiving further quality information about a further coating profile from an artificial intelligence and receiving image data about the further coating profile from an artificial intelligence.
  • the database can be expanded in this way and can be continuously improved with the use of artificial intelligence.
  • the coating profile can be a coating profile of a raceway in a cylinder crankcase and the image data can be a micrograph of the Include coating profile.
  • a database can be created during the quality control in the manufacture of cylinder crankcases.
  • Embodiments also provide a device for generating quality information about a coating profile.
  • the device comprises at least one interface for communicating image data and a control module which is designed to carry out one of the methods described herein for generating quality information about a coating profile.
  • Another exemplary embodiment is a monitoring device for quality control in the manufacture of cylinder crankcases with a device for generating quality information about a coating profile.
  • a device for generating a database with quality information about coating profiles is a further exemplary embodiment.
  • the device comprises at least one interface for communicating image data and a control module for performing one of the methods described herein for generating a database with quality information about coating profiles.
  • Another exemplary embodiment is a computer program for carrying out a method described herein when the computer program runs on a computer, a processor, a control module or a programmable hardware component.
  • FIG. 1 shows a block diagram of a flow chart of an exemplary embodiment of a method for generating quality information about a coating profile
  • FIG. 2 shows a block diagram of a flowchart of an exemplary embodiment of a method for generating a database with quality information about coating profiles
  • FIG. 3 shows a block diagram of an exemplary embodiment of a device for generating quality information about a coating profile and / or for generating a database with quality information about coating profiles;
  • FIG. 4 shows a process flow with quality control in an exemplary embodiment
  • 5 shows a diagram for explaining atmospheric plasma spraying in an exemplary embodiment
  • FIG. 8 shows further image data of coating profiles for evaluation in an exemplary embodiment.
  • the method 10 comprises a sensory acquisition 12 of image data about the coating profile and an evaluation 14 of the image data using artificial intelligence in order to obtain the quality information about the coating profile.
  • the method 10 can further comprise training the artificial intelligence based on a database with quality information about coating profiles.
  • a coating profile is understood here to mean a cross section that shows at least two layers and their transition.
  • a layer can be assessed with regard to its layer thickness and regularity and the transition to one or more neighboring layers.
  • a coating profile can be generated by a cut or polished image.
  • a corresponding workpiece is separated and the separating surface is ground.
  • Image data can then be generated from the ground separating surface, e.g. by photography, which then result in a micrograph. All common methods for generating image data are conceivable for the sensory acquisition of the image data. Examples are detection by means of a laser, an X-ray device and / or an optical detection method (photography).
  • the artificial intelligence can be used, for example, in the form of a neural network that is first trained with a database and then used for automated image evaluation. However, it is conceivable to use any machine learning algorithms, some of which are explained below.
  • the methods 10 described herein can therefore be based on a machine learning model or a machine learning algorithm.
  • the artificial intelligence can be further developed, for example by the fact that the method 10 also includes a training of the artificial intelligence based on an assessment of the quality information by a user. Assessments by Experts can then be used to reduce false assessments caused by artificial intelligence or, ideally, even to avoid them entirely.
  • FIG. 2 shows a block diagram of a flowchart of an exemplary embodiment of a method 20 for generating a database with quality information about coating profiles.
  • the method comprises a sensor-based acquisition 22 of image data about the coating profile and an evaluation 24 of the image data by a user in order to obtain the quality information about the coating profiles.
  • the method 20 further comprises storing 26 the quality information about the coating profiles in the database.
  • a database generated in this way can then be used to train an artificial intelligence.
  • FIG. 3 shows a block diagram of an exemplary embodiment of a device 30, 40 for generating quality information about a coating profile and / or for generating a database with quality information about coating profiles.
  • the device 30 for generating quality information about the coating profile comprises at least one interface 32 for communicating image data and a control module 34 which is coupled to the at least one interface and is designed to control the at least one interface 32.
  • the control module 32 is also designed to carry out one of the methods 10 described herein for generating quality information about the coating profile.
  • 3 also illustrates, as optional components (dashed line), a monitoring device 100 for quality control in the manufacture of cylinder crankcases with a device 30.
  • the 3 also shows an exemplary embodiment of a device 40 for generating a database with quality information about coating profiles.
  • the device 40 comprises at least one interface 42 for communication of image data, which is coupled to a control module 44.
  • the control module 44 is designed to control the at least one interface 42 and to carry out at least one of the methods 20 described herein.
  • the at least one interface 32, 42 of the device 30, 40 can, in exemplary embodiments, be designed in the form of contacts for wired communication or as antennas for cordless communication. In exemplary embodiments, they can also be designed as separate hardware. They can include memories that at least temporarily store the signals to be sent or received.
  • the at least one interface 32, 42 can be designed to receive electrical signals be, for example, as a bus interface, or as an optical interface. In addition, it can be designed for radio transmission in exemplary embodiments and comprise a radio front end and associated antennas. Furthermore, the at least one interface 32, 42 can comprise synchronization mechanisms for synchronization with the respective transmission medium. In exemplary embodiments, the at least one interface 32, 42 can be designed to communicate with corresponding sensors or sensory acquisition components in order to receive or acquire at least the image data.
  • control module 34, 44 can have any processor cores, such as digital signal processor cores (DSPs). Embodiments are not restricted to a specific type of processor core. Any processor cores or also several processor cores or microcontrollers for implementing the control module 34, 44 are conceivable. Implementations in integrated form with other devices are also conceivable, for example in a control unit for a system with further components which additionally include one or more other functions.
  • DSPs digital signal processor cores
  • the control module 34, 44 can therefore correspond to any component that can carry out a corresponding evaluation using artificial intelligence.
  • exemplary embodiments are explained using the example of APS layers. Such coating processes are in principle in use worldwide.
  • software solutions can be used to support the evaluation and the creation of databases, e.g. pixel colorers. In some exemplary embodiments, these can be provided to support a user or else they can be automated.
  • the artificial intelligence can then be used for the automated evaluation of micrographs and / or layers. This enables a higher tolerance to preparation errors, artifacts, environmental conditions (e.g. different lighting), influence of the inspector, etc. to be achieved.
  • a central Database / database can be generated, which makes the evaluations of, for example, more experienced locations available to less experienced locations.
  • the quality of APS layers can be assessed on the basis of manual evaluation of micrographs in an image database, the manual evaluation being very time-consuming.
  • the criteria to be evaluated are described, for example, in a corresponding specification (such as a workshop sketch) and are converted into test regulations.
  • a corresponding quality control across locations can still be inconsistent, since the results are strongly influenced by the recording quality, the examiner and the experience of the evaluator. This means that the results of different auditors cannot be compared.
  • image evaluation software can be used that performs automated image recording and image evaluation, independently of the examiner.
  • a location-independent evaluation routine can thus be implemented, which can offer an automation of the evaluation, in particular for series monitoring, and thus a significant potential for efficiency.
  • Embodiments can thus provide a uniform evaluation standard for APS coatings.
  • an implementation independent of the examiner can take place and, if necessary, a significant increase in efficiency in the analysis and evaluation of APS layers can be achieved through automation.
  • APS atmospheric plasma spraying
  • the layer represents the partner of the piston rings and significantly influences wear and oil consumption.
  • the cylinder crankcase is first mechanically prepared 401, and the associated quality check is a check of the dimensional accuracy 402.
  • the surface that is to be coated is then laser roughened 403. The roughness can be checked, for example, using a confocal optical method.
  • the coating is then thermally sprayed on 405 and the layer adhesion is checked by means of a positive test (adhesion test) 406. This is followed by pre-honing 407 with eddy current testing 408 and finish honing (smooth honing) 409 with confocal honing evaluation 410.
  • the steps 403 and 405 can also be checked by a metallographic section analysis.
  • a random sample of the workpieces from series production is taken and separated accordingly.
  • the workpiece is, for example, placed in an embedding compound (eg plastic compound) that delimits the cut surface and a Prevents the formation of burrs or the rounding of edges.
  • the parting surface is then ground and recorded using sensors, e.g. photographed, in order to generate image data of the parting surface (coating profile).
  • an enlarged representation eg microscopic
  • the coating profile is then a coating profile of the raceway in the cylinder crankcase, and the image data correspond to a micrograph of the coating profile. These can then be evaluated automatically in exemplary embodiments.
  • Fig. 5 is a diagram for explaining the atmospheric plasma spray in the embodiment. 5 shows a spray nozzle with an inner electrode 501 and an outer electrode 502 (actual nozzle) between which a voltage 503 is applied. Inner and outer electrodes are insulated from one another 504, and an arc occurs between electrodes 501, 502. Plasma gas 505 is introduced under high pressure and flows out at high temperature through the arc and nozzle 502 to the right in FIG. At the same time, metal particles are introduced 506 via a carrier gas into the outflowing plasma arc, where they melt and a spray jet 507 is created. This applies a coating 508 to the substrate 509. Coolant 510 can be introduced to cool the apparatus.
  • FIG. 5 illustrates an APS method in which a metallic layer 508 is sprayed onto a roughened surface 509.
  • An arc is generated between an anode (nozzle) 502 and a cathode 501 (electrode) for melting.
  • argon gases
  • 506 entering the device with the generated arc, a high thermal energy is released. This thermal energy is used to melt the metal powders 506 and accelerate them for spraying.
  • the evaluation of the image data can include an assessment of at least one element from the group of a layer thickness, a layer transition, a defect, a pore, a cavity, a course of a layer transition, an oxidation and a foreign material.
  • OK OK
  • NOK not OK
  • the evaluation can also provide for a distinction between pore proportions and oxide proportions, the evaluation being heavily dependent on the image quality.
  • the investment material can have different colors, examples are transparent, green, blue, black, mixed, etc.
  • the surface of the layer may have been processed differently, for example the interface to the investment material is smooth (finished) or rough (unprocessed).
  • FIG. 6 shows image data of coating profiles for evaluation in an exemplary embodiment.
  • Fig. 6 shows on the left a coating profile with three layers, the investment material can be seen above, the APS layer below and the roughened carrier profile (base material aluminum with pores) can be seen below.
  • the APS layer shows oxides (gray) in the coating, which make up a certain percentage, and darker (black) pores, which also make up a certain area in the coating.
  • globular particles can be made out, a certain number of which can be observed per image section.
  • Another image of the coating profile is shown on the right in FIG. 6.
  • the laser roughening of the carrier surface can be seen here, with the distance between the "mountains” in the roughened profile and their heights (46 pm) being an assessment criterion.
  • the number of “mountains” per image section and the height of the individual “mountains” can also be evaluated to form a dynamic reference line 600 in such an image section.
  • preparation artifacts and defects can be seen in FIG. 6.
  • FIG. 7 shows image data of coating profiles for evaluation in an exemplary embodiment with a coating detachment and pores.
  • FIG. 7 shows a coating profile with three layers, the embedding compound being visible above, the APS layer underneath and the roughened carrier profile below.
  • Fig. 7 shows one on the left Gap between the coating and the aluminum carrier, and defects in the structure of the roughening can also be seen.
  • the profile on the left-hand side of FIG. 7 is not in order due to the delamination and would not pass the quality control. In addition, a further investigation would take place in this case in order to get to the bottom of the cause of the deficient roughening and the delamination.
  • FIG. 7 shows an image with the same coatings, this coating profile passing the quality control and being found to be in order, although layer defects 700 can also be seen here.
  • FIG. 8 illustrates further image data of coating profiles for evaluation in an exemplary embodiment.
  • a coating profile with inadequate laser roughening can be seen, with hardly any profiling visible at the transition from carrier to coating. This workpiece would not pass the quality control and the laser roughening would be checked or readjusted in the manufacturing process.
  • FIG. 8 shows a coating profile with residues (detached particles) from the laser roughening in the coating, which leads to layer defects.
  • the coating profile on the right is borderline but would pass quality control.
  • FIGS. 6, 7 and 8 show how the quality of the coating profile can be assessed on the basis of the criteria described.
  • Such examples can be stored in the database and serve as a basis for training the artificial intelligence.
  • the quality information about the coating profile can be stored in the database.
  • the database can be expanded and expanded with examples that have already been assessed by the artificial intelligence.
  • the image data can also be stored in the database via the coating profile, so that training can be carried out across locations with the same growing databases. An exchange between various such methods 10 can also be implemented.
  • a further method 10 can also include receiving further quality information about a further coating profile from another artificial intelligence and receiving image data about the further coating profile from the other artificial intelligence.
  • Embodiments can be based on the use of a machine learning model or machine learning algorithm.
  • Machine learning can refer to algorithms and statistical models that computer systems can use to perform a particular task without using explicit instructions, rather than relying on models and inference.
  • a transformation of data can be used that can be derived from an analysis of course and / or training data (database described above).
  • the content of images can be analyzed using a machine learning model or using a machine learning algorithm. So that the machine learning model can analyze the content of an image, the machine learning model can be trained using training images as input and training content information as output.
  • the machine learning model “learns” the content of the images so that the content of images that are not included in the training data can be recognized using the machine learning model.
  • the same principle can be used for other types of sensor data as well: By training a machine learning model using training sensor data and a desired output, the machine learning model “learns” a conversion between the sensor data and the output, which can be used to create a To provide output based on non-training sensor data provided to the machine learning model.
  • the data provided (for example sensor data, metadata and / or image data) can be preprocessed in order to obtain a feature vector which is used as an input for the machine learning model.
  • Machine learning models can be trained using training input data.
  • the above examples use a training process called “supervised learning”.
  • supervised learning the machine learning model is trained using a plurality of training samples, each sample being able to include a plurality of input data values and a plurality of desired output values, ie a desired output value is assigned to each training sample.
  • the machine learning model "learns" which output value (coating profile okay or not okay) based on an input sample value (image data or preprocessed image data), which is similar to the samples provided during training.
  • semi-supervised learning can also be used.
  • Supervised learning can be based on a supervised learning algorithm (e.g. a classification algorithm, a regression algorithm, or a similarity learning algorithm).
  • Classification algorithms can be used when the outputs are constrained to a limited set of values (categorical variables); H. the input is classified as one of the limited set of values.
  • Regression algorithms can be used when the outputs show any numerical value (within a range).
  • Similarity learning algorithms can be similar to both classification and regression algorithms, but are based on learning from examples using a similarity function that measures how similar or related two objects are.
  • unsupervised learning can be used to train the machine learning model.
  • unsupervised learning (only) input data may be provided and an unsupervised learning algorithm can be used to find a structure in the input data (e.g. by grouping or clustering the input data, finding similarities / correlations in the data) .
  • Clustering is the assignment of input data that comprise a plurality of input values in subsets (clusters) so that input values within the same cluster are similar according to one or more (predefined) similarity criteria, while they are dissimilar to input values that are included in other clusters.
  • Reinforcement learning is a third group of machine learning algorithms.
  • reinforcement learning can be used to train the machine learning model.
  • one or more software actors are trained to take action in an environment.
  • a reward is calculated based on the actions taken.
  • Reinforcement learning is based on training the one or more software agents to select the actions to increase the cumulative reward, resulting in software agents who become better at the task they are given (such as increasing rewards proven).
  • feature learning can be used.
  • that Machine learning model can be trained at least in part using feature learning, and / or the machine learning algorithm can comprise a feature learning component.
  • Feature learning algorithms called representation learning algorithms, can preserve the information in their input but transform it so that it becomes useful, often as a preprocessing stage before performing the classification or prediction.
  • feature learning can be based on a principal component analysis or a cluster analysis.
  • anomaly detection ie, outlier detection, e.g. gap in FIG. 7 on the left, residue in FIG. 8 on the right
  • the machine learning model can be trained at least in part using anomaly detection and / or the machine learning algorithm can include an anomaly detection component.
  • the machine learning algorithm can use a decision tree as a predictive model.
  • the machine learning model can be based on a decision tree.
  • the observations on an item e.g., a set of input values
  • an output value corresponding to the item can be represented by the leaves of the decision tree.
  • Decision trees can support both discrete values and continuous values as output values. If discrete values are used, the decision tree can be called a classification tree, if continuous values are used the decision tree can be called a regression tree.
  • Association rules are another technique that can be used in machine learning algorithms.
  • the machine learning model can be based on one or more association rules.
  • Association rules are created by identifying relationships between variables in large amounts of data.
  • the machine learning algorithm can identify and / or use one or more relationship rules that represent the knowledge derived from the data.
  • the rules can e.g. B. used to store, manipulate or apply the knowledge.
  • Machine learning algorithms are usually based on a machine learning model.
  • the term “machine learning algorithm” can refer to a set of instructions that can be used to create, train, or use a machine learning model.
  • the term “machine learning model” can denote a data structure and / or a set of rules that represents the knowledge learned (e.g.
  • the use of a machine learning algorithm can imply the use of an underlying machine learning model (or a plurality of underlying machine learning models).
  • the use of a machine learning model can imply that the machine learning model and / or the data structure / set of rules which the machine learning model is / are is trained by a machine learning algorithm.
  • the machine learning model can be an artificial neural network (ANN).
  • ANNs are systems inspired by biological neural networks such as those found in a retina or a brain.
  • ANNs comprise a plurality of interconnected nodes and a plurality of connections, so-called edges, between the nodes.
  • Each node can represent an artificial neuron.
  • Each edge can send information from one node to another.
  • the output of a node can be defined as a (nonlinear) function of the inputs (e.g. the sum of its inputs).
  • a node's inputs can be used in the function based on a "weight" of the edge or the node providing the input.
  • the weight of nodes and / or of edges can be adjusted in the learning process.
  • the training of an artificial neural network can comprise an adaptation of the weights of the nodes and / or edges of the artificial neural network, i. E. H. to achieve a desired output for a particular input.
  • the machine learning model can be a support vector machine, a random forest model or a gradient boosting model.
  • Support vector machines are supervised learning models with associated learning algorithms that can be used to analyze data (e.g. in a classification or regression analysis).
  • Support vector machines can be trained by providing input with a plurality of training input values belonging to one of two categories.
  • the Support Vector Machine can be trained to one of two categories assign a new input value.
  • the machine learning model can be a Bayesian network that is a probabilistic directional acyclic graphical model.
  • a Bayesian network can represent a set of random variables and their conditional dependencies using a directed acyclic graph.
  • the machine learning model can be based on a genetic algorithm, which is a search algorithm and heuristic technique that mimics the process of natural selection.
  • Exemplary embodiments can furthermore be a computer program with a program code for executing one or more of the above methods or refer to them when the computer program is executed on a computer or processor. Steps, operations or processes of various methods described above can be carried out by programmed computers or processors. Examples can also include program storage devices, e.g. Digital data storage media that are machine, processor, or computer readable and encode machine, processor, or computer executable programs of instructions. The instructions perform or cause some or all of the steps in the procedures described above.
  • the program storage devices may e.g. B. digital storage, magnetic storage media such as magnetic disks and tapes, hard disk drives or optically readable digital data storage media or be.
  • Functions of various elements shown in the figures as well as the designated function blocks can be in the form of dedicated hardware, e.g. B “a signal provider”, “a signal processing unit”, “a processor”, “a controller” etc. as well as being implemented as hardware capable of executing software in conjunction with the associated software.
  • the functions can be provided by a single dedicated processor, by a single shared processor, or by a plurality of individual processors, some or all of which can be shared.
  • processor or “controller” is by no means limited to hardware that is exclusively capable of executing software, but can also include digital signal processor hardware (DSP hardware;
  • DSP Digital Signal Processor
  • network processor application-specific integrated circuit Application Specific Integrated Circuit (ASIC), Field Programmable Gate Array (FPGA), Read Only Memory (ROM) for storing software, Random Access Memory (RAM), and non-volatile storage device.
  • ASIC Application Specific Integrated Circuit
  • FPGA Field Programmable Gate Array
  • ROM Read Only Memory
  • RAM Random Access Memory
  • non-volatile storage device Other hardware, conventional and / or custom, can also be included.
  • a block diagram may represent a high level circuit diagram that implements the principles of the disclosure.
  • a flowchart, sequence diagram, state transition diagram, pseudocode, and the like may represent various processes, operations, or steps, for example, essentially represented in computer-readable medium and thus performed by a computer or processor, whether or not such Computer or processor is shown explicitly.
  • Methods disclosed in the description or in the claims can be implemented by a device having a means for performing each of the respective steps of these methods.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

Des modes de réalisation de l'invention concernent un procédé, un programme informatique et un dispositif de génération d'informations de qualité, un procédé, un programme informatique et un dispositif de génération de base de données, et un dispositif de surveillance. Le procédé (10) de génération d'informations de qualité concernant un profil de revêtement comprend l'acquisition basée sur un capteur (12) de données d'image concernant le profil de revêtement et l'évaluation (14) des données d'image à l'aide d'une intelligence artificielle pour obtenir les informations de qualité concernant le profil de revêtement.
PCT/EP2021/060712 2020-04-29 2021-04-23 Procédé, dispositif et programme informatique de génération d'informations de qualité concernant un profil de revêtement, procédé, dispositif et programme informatique de génération de base de données, et dispositif de surveillance WO2021219515A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE102020205456.8A DE102020205456A1 (de) 2020-04-29 2020-04-29 Verfahren, Vorrichtung und Computerprogramm zum Erzeugen von Qualitätsinformation über ein Beschichtungsprofil, Verfahren, Vorrichtung und Computerprogramm zum Erzeugen einer Datenbank, Überwachungsgerät
DEDE102020205456.8 2020-04-29

Publications (1)

Publication Number Publication Date
WO2021219515A1 true WO2021219515A1 (fr) 2021-11-04

Family

ID=75690288

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2021/060712 WO2021219515A1 (fr) 2020-04-29 2021-04-23 Procédé, dispositif et programme informatique de génération d'informations de qualité concernant un profil de revêtement, procédé, dispositif et programme informatique de génération de base de données, et dispositif de surveillance

Country Status (2)

Country Link
DE (1) DE102020205456A1 (fr)
WO (1) WO2021219515A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114660076A (zh) * 2022-05-19 2022-06-24 张家港市欧凯医疗器械有限公司 一种医疗用管涂层质量检测方法及系统
DE102022213893A1 (de) 2022-12-19 2024-06-20 Siemens Aktiengesellschaft System und Verfahren zur Bereitstellung eines Lackierprogramms für die Beschichtung von Leiterplatten

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116912243B (zh) * 2023-09-12 2023-11-21 吉林大学第一医院 一种用于医疗用管的涂层质量检测方法及系统

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180342069A1 (en) * 2017-05-25 2018-11-29 General Electric Company Neural network feature recognition system
WO2019099928A2 (fr) 2017-11-17 2019-05-23 Kevin Friesth Système de fabrication automatisé avancé et procédés pour composants thermiques et mécaniques utilisant un frittage laser direct hybride quadratique ou au carré, frittage laser métallique direct, commande numérique par ordinateur, pulvérisation thermique, dépôt de métal direct et soudage par friction-malaxage
DE102018128478A1 (de) 2017-11-28 2019-05-29 General Electric Company Adaptive Roboterzelle zur Thermosprühbeschichtung

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102008048109B4 (de) 2008-04-17 2015-01-29 Ks Aluminium-Technologie Gmbh Verfahren zur Herstellung eines metallischen Bauteils und Verwendung eines Zylinderteils als Grundkörper zur Durchführung des Verfahrens
FR3074596B1 (fr) 2017-12-01 2019-12-06 Universite De Reims Champagne Ardenne Procede de caracterisation d'echantillons utilisant des reseaux de neurones

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180342069A1 (en) * 2017-05-25 2018-11-29 General Electric Company Neural network feature recognition system
WO2019099928A2 (fr) 2017-11-17 2019-05-23 Kevin Friesth Système de fabrication automatisé avancé et procédés pour composants thermiques et mécaniques utilisant un frittage laser direct hybride quadratique ou au carré, frittage laser métallique direct, commande numérique par ordinateur, pulvérisation thermique, dépôt de métal direct et soudage par friction-malaxage
DE102018128478A1 (de) 2017-11-28 2019-05-29 General Electric Company Adaptive Roboterzelle zur Thermosprühbeschichtung

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
BADMOS OLATOMIWA ET AL: "Image-based defect detection in lithium-ion battery electrode using convolutional neural networks", JOURNAL OF INTELLIGENT MANUFACTURING, CHAPMAN AND HALL, LONDON, GB, vol. 31, no. 4, 1 August 2019 (2019-08-01), pages 885 - 897, XP037083271, ISSN: 0956-5515, [retrieved on 20190801], DOI: 10.1007/S10845-019-01484-X *
CHEN WEI-BANG ET AL: "A Fully Automated Porosity Measure for Thermal Barrier Coating Images :", INTERNATIONAL JOURNAL OF MULTIMEDIA DATA ENGINEERING AND MANAGEMENT (IJMDEM), vol. 9, no. 4, 1 October 2018 (2018-10-01), US, pages 40 - 58, XP055823500, ISSN: 1947-8534, DOI: 10.4018/IJMDEM.2018100103 *
CHEN WEI-BANG ET AL: "Automatic Classification of Microstructures in Thermal Barrier Coating Images", 2017 IEEE INTERNATIONAL SYMPOSIUM ON MULTIMEDIA (ISM), IEEE, 11 December 2017 (2017-12-11), pages 99 - 106, XP033291465, DOI: 10.1109/ISM.2017.119 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114660076A (zh) * 2022-05-19 2022-06-24 张家港市欧凯医疗器械有限公司 一种医疗用管涂层质量检测方法及系统
DE102022213893A1 (de) 2022-12-19 2024-06-20 Siemens Aktiengesellschaft System und Verfahren zur Bereitstellung eines Lackierprogramms für die Beschichtung von Leiterplatten

Also Published As

Publication number Publication date
DE102020205456A1 (de) 2021-11-04

Similar Documents

Publication Publication Date Title
WO2021219515A1 (fr) Procédé, dispositif et programme informatique de génération d'informations de qualité concernant un profil de revêtement, procédé, dispositif et programme informatique de génération de base de données, et dispositif de surveillance
EP3432266B1 (fr) Procédé et dispositif de détection de défauts
EP3767403B1 (fr) Mesure de forme et de surface assistée par apprentissage automatique destinée à la surveillance de production
DE102019110721A1 (de) Workflow zum trainieren eines klassifikators für die qualitätsprüfung in der messtechnik
DE112017005651T5 (de) Vorrichtung zur Klassifizierung von Daten
EP3970113B1 (fr) Détermination du niveau d'usure d'un outil
EP3671632A1 (fr) Prévision d'entretien à base d'image et détection de fausses man uvres
DE102018109819A1 (de) Verfahren zur Gewinnung von Information aus Röntgen-Computertomographiedaten zur Optimierung des Spritzgussprozesses von kurzfaserverstärkten Kunststoffteilen
EP3479178B1 (fr) Regulation d'un systeme de production basé sur une simulation
DE102019217613A1 (de) Verfahren zur diagnose eines motorzustands und diagnostisches modellierungsverfahren dafür
CN112017204A (zh) 一种基于边缘标记图神经网络的刀具状态图像分类方法
DE102021201124A1 (de) Trainieren von bildklassifizierernetzen
EP3635680A1 (fr) Procédé et dispositif mis en oeuvre par un ordinateur pour générer automatiquement des données d'image caractérisées et dispositif d'analyse pour contrôler un composant
DE102019117680A1 (de) Automatisiertes segmentierungsverfahren für die qualitätsprüfung
DE102018220608A1 (de) Maschinelles Lernsystem, sowie ein Verfahren, ein Computerprogramm und eine Vorrichtung zum Erstellen des maschinellen Lernsystems
DE102022203475A1 (de) System zum Erzeugen einer von einem Menschen wahrnehmbaren Erklärungsausgabe für eine von einem Anomalieerkennungsmodul vorhergesagte Anomalie auf hochfrequenten Sensordaten oder davon abgeleiteten Größen eines industriellen Fertigungsprozesses, Verfahren und Computerprogramm zur Überwachung einer auf künstlicher Intelligenz basierenden Anomalieerkennung in hochfrequenten Sensordaten oder davon abgeleiteten Größen eines industriellen Fertigungsprozesses und Verfahren und Computerprogramm zur Überwachung einer auf künstlicher Intelligenz basierenden Anomalieerkennung bei einer End-of-Line Akustikprüfung eines Getriebes
EP4095748A1 (fr) Fabrication d'une installation d'inspection de qualité
DE102020212510A1 (de) Verfahren und Vorrichtung zum Aufzeigen des Einflusses von Schneidparametern auf eine Schnittkante
EP4094867A1 (fr) Procédé de fabrication additive d'un composant
EP3605404B1 (fr) Procédé et dispositif d'entraînement d'une routine d'apprentissage mécanique permettant de commander un système technique
DE10013137B4 (de) Verfahren zur bildgesteuerten Prüfung und Bearbeitung von Produkten
DE102018109816B3 (de) Verfahren zur Gewinnung mindestens eines signifikanten Merkmals in einer Serie von Bauteilen gleichen Typs und Verfahren zur Klassifikation eines Bauteils eienr solchen Serie
DE102020212005B4 (de) Verfahren zur Bestimmung der für einen Bildklassifikator entscheidungsrelevanten Bildanteile
WO2022037899A1 (fr) Procédé et appareil de fabrication additive d'une pièce à travailler
DE102022103844B3 (de) Verfahren zur Optimierung eines Produktionsprozesses auf Basis visueller Informationen und Vorrichtung zur Durchführung des Verfahrens

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21721906

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21721906

Country of ref document: EP

Kind code of ref document: A1