WO2023078747A1 - System for training a deep-learning algorithm and associated method - Google Patents

System for training a deep-learning algorithm and associated method Download PDF

Info

Publication number
WO2023078747A1
WO2023078747A1 PCT/EP2022/079878 EP2022079878W WO2023078747A1 WO 2023078747 A1 WO2023078747 A1 WO 2023078747A1 EP 2022079878 W EP2022079878 W EP 2022079878W WO 2023078747 A1 WO2023078747 A1 WO 2023078747A1
Authority
WO
WIPO (PCT)
Prior art keywords
learning algorithm
image
images
deep
classification
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/EP2022/079878
Other languages
French (fr)
Inventor
Herve Carrerot
Yoann HEBRARD
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
SKF Aerospace France SAS
SKF AB
Original Assignee
SKF Aerospace France SAS
SKF AB
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 SKF Aerospace France SAS, SKF AB filed Critical SKF Aerospace France SAS
Priority to CN202280074107.3A priority Critical patent/CN118302800A/en
Priority to KR1020247018217A priority patent/KR20240118084A/en
Priority to GB2405441.3A priority patent/GB2625498A/en
Priority to JP2024525899A priority patent/JP2024543023A/en
Priority to CA3236753A priority patent/CA3236753A1/en
Priority to US18/706,495 priority patent/US20250005734A1/en
Priority to DE112022005358.5T priority patent/DE112022005358T5/en
Publication of WO2023078747A1 publication Critical patent/WO2023078747A1/en
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/809—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of classification results, e.g. where the classifiers operate on the same input data
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16C—SHAFTS; FLEXIBLE SHAFTS; ELEMENTS OR CRANKSHAFT MECHANISMS; ROTARY BODIES OTHER THAN GEARING ELEMENTS; BEARINGS
    • F16C19/00—Bearings with rolling contact, for exclusively rotary movement
    • F16C19/52—Bearings with rolling contact, for exclusively rotary movement with devices affected by abnormal or undesired conditions
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16C—SHAFTS; FLEXIBLE SHAFTS; ELEMENTS OR CRANKSHAFT MECHANISMS; ROTARY BODIES OTHER THAN GEARING ELEMENTS; BEARINGS
    • F16C33/00—Parts of bearings; Special methods for making bearings or parts thereof
    • F16C33/30—Parts of ball or roller bearings
    • F16C33/32—Balls
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16C—SHAFTS; FLEXIBLE SHAFTS; ELEMENTS OR CRANKSHAFT MECHANISMS; ROTARY BODIES OTHER THAN GEARING ELEMENTS; BEARINGS
    • F16C33/00—Parts of bearings; Special methods for making bearings or parts thereof
    • F16C33/30—Parts of ball or roller bearings
    • F16C33/34—Rollers; Needles
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/10—Image acquisition
    • G06V10/12—Details of acquisition arrangements; Constructional details thereof
    • G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
    • G06V10/143—Sensing or illuminating at different wavelengths
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • G06V10/449—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
    • G06V10/451—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters with interaction between the filter responses, e.g. cortical complex cells
    • G06V10/454—Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/778—Active pattern-learning, e.g. online learning of image or video features
    • G06V10/7784—Active pattern-learning, e.g. online learning of image or video features based on feedback from supervisors
    • G06V10/7792—Active pattern-learning, e.g. online learning of image or video features based on feedback from supervisors the supervisor being an automated module, e.g. "intelligent oracle"
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16C—SHAFTS; FLEXIBLE SHAFTS; ELEMENTS OR CRANKSHAFT MECHANISMS; ROTARY BODIES OTHER THAN GEARING ELEMENTS; BEARINGS
    • F16C2206/00—Materials with ceramics, cermets, hard carbon or similar non-metallic hard materials as main constituents
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/06—Recognition of objects for industrial automation

Definitions

  • the present invention relates to ceram ic rolling elements, and more particularly relates to the detection of manufacturing defects in such rolling elements.
  • rolling bearings equipped with rolling elements made of ceram ic or of steel.
  • the rolling elements may for example be balls or even cylindrical, conical or spherical rollers.
  • rolling elements of rolling bearings allow a circular movement of a shaft with respect to a fixed element to be ensured while lim iting frictional force.
  • ceramic rolling elements are produced by sintering, they may contain inclusions of foreign material, non-uniform agglomerations of material, or porosities.
  • the rolling bearing equipped with failed rolling elements when they is located on a rotating shaft, they may cause spalling and/or overheating and the shaft will rotate with increasing difficulty.
  • the rolling elements are liable to completely disunite from the other components of the bearing and thus cause the rotating shaft to disassociate from its surrounding mechanical system : this may prove to be critical in aeronautics.
  • the defectrecognition procedure may slow the entirety of the rolling-element production line.
  • the operator may make m istakes in his interpretation of the images.
  • the goal of the invention is to overcome the aforementioned constraints.
  • One subject of the invention is therefore a method for training a deeplearning algorithm for detecting a defect in a ceram ic rolling element, comprising:
  • the analysis of the data of the classification performed by the learning algorithm makes it possible to interpret the variation in the weights of the deeplearning algorithm when it comprises a neural network.
  • the procedure for recognition of defects in the rolling element may be automated so as to increase the production rate of a line for producing rolling elements while m inimizing the risks of error in the interpretation of the image set.
  • the generation of a training data set comprises: -segmenting each image into various regions, -filtering a second time each region of each image of the set of images to obtain a data vector, the data vector forming the training data set.
  • the classification of each image of the set of radiographic images by a statistical learning algorithm comprises:
  • the image-analysis indicator comprises computing an average, and/or a standard deviation, and/or a median, and/or a greyscale gradient.
  • the training of a deep-learning algorithm comprises: -segmenting each filtered image into various regions, and -processing each region of each image using the deep-learning algorithm with a view to detecting at least one defect on the basis of a selection criterion, -detecting a defect in light of a selection criterion specific to each region, and
  • the selection criterion comprises the geometry of the rolling element in said region, and/or texture in said region and/or contrast in said region.
  • the deep-learning algorithm comprises a convolutional neural network.
  • the comparison of the classification of the images of the image set performed by the statistical learning algorithm and of the classification of the images of the image set performed by the deep-learning algorithm comprises:
  • Another subject of the invention is a system for training a deep-learning algorithm for detecting a defect in a ceram ic rolling element, comprising:
  • -filtering means configured to improve the contrast of each image of the set of images
  • -implementing means configured to implement a statistical learning algorithm configured to classify, on the basis of the data set, each image of the image set into a class of suspect rolling elements or into a class of non-suspect rolling elements,
  • -training means configured to train, on the basis of the set of radiographic images, a deep-learning algorithm so that the deep-learning algorithm classifies each image of the set of radiographic images into the class of suspect rolling elements or into the class of non-suspect rolling elements, and
  • -comparing means configured to compare the classification of the images of the image set performed by the statistical learning algorithm and the classification of the images of the image set performed by the deep-learning algorithm with a view to determ ining the accuracy of the classification of the deep-learning algorithm with respect to the classification performed by the statistical learning algorithm .
  • FIG. 1 illustrates a system for training a deep-learning algorithm for detecting a defect in a ceram ic rolling element, according to one embodiment of the invention
  • FIG 2 shows one mode of implementation of the training system according to the invention
  • FIG. 3 shows one example of a filtered image according to the invention
  • Fig 4 shows one example of division of a filtered image
  • FIG 5 shows one example of a data set according to the invention. Detailed description of the invention
  • Figure 1 shows a system for training a deep-learning algorithm ALG1 for detecting a defect in a ceram ic rolling element 2.
  • the system 1 comprises means 3 for capturing a set of two-dimensional digital radiographic images DATA1 of rolling elements 2 comprising defects and not comprising any defects,
  • the capturing means 3 comprise a photographic sensor 4 and a generator 5 of high-frequency electromagnetic waves R1 configured to em it the electromagnetic waves R1 towards the element 2.
  • a two-dimensional digital radiographic image of the rolling element 2 is thus formed.
  • the generator 4 generally takes the form of a m icrofocus x-ray tube.
  • the capturing means 3 generate a three-dimensional digital radiographic image.
  • the system 1 further comprises filtering means 6 that are connected to the capturing means 3 and that process the images of the set of images to improve the contrast of each image of the set of images DATA1 captured by the capturing means 3.
  • the filtering means 6 deliver a set of filtered radiographic images DATA2.
  • the system 1 also comprises means 7 for generating a data set DATA3 from the set of radiographic images DATA2 filtered by the filtering means 6, and implementing means 8 that implement a statistical learning algorithm ALGO1 that is configured to classify, on the basis of the data set DATA3, each image of the set of filtered images DATA2 into a class CL1 of suspect rolling elements 2 or into a class CL2 of non-suspect rolling elements 2.
  • the generating means 7 are connected to the filtering means 6 and to the implementing means 8.
  • the classification performed by the statistical learning algorithm ALGO1 is for example stored in a memory 9.
  • the statistical learning algorithm ALGO1 for example comprises a decision-tree algorithm , a random-forest algorithm , a support-vector-machine algorithm , a K-nearest-neighbours algorithm , or a logistic-regression algorithm .
  • a suspect rolling element is a rolling element 2 the probability of having at least one defect of which is higher than its probability of not having any defects.
  • Defects for example comprise inclusions of foreign material, non-uniform agglomerations of material or porosities.
  • the system 1 further comprises training means 10 that train, on the basis of the set of filtered radiographic images DATA2, a deep-learning algorithm ALGO2 so that the deep-learning algorithm classifies each image of the set of filtered radiographic images DATA2 into the class CL1 of suspect rolling elements or into the class CL2 of non-suspect rolling elements.
  • the training means 10 are connected to the filtering means 6.
  • the deep-learning algorithm ALGO2 for example comprises a convolutional neural network.
  • the classification performed by the deep-learning algorithm ALGO2 is for example stored in a memory 1 1 .
  • the system 1 further comprises comparing means 12 that compare the classification performed by the statistical learning algorithm ALGO1 , and the classification performed by the deep-learning algorithm ALGO2 in order to determ ine the accuracy of the deep-learning algorithm on the basis of the classification of the statistical learning algorithm .
  • the comparing means 12 are connected to the implementing means 8 and to the training means 10, and for example read the content of the memory 9 of the implementing means storing the classification performed by the statistical learning algorithm ALGO1 and the content of the memory 1 1 of the training means 10 storing the classification performed by the deep-learning algorithm ALGO2.
  • the system 1 further comprises a processing unit UT that employs the capturing means 3 to capture the set of images DATA1 , the filtering means, means 7 for generating a data set DATA3 on the basis of the set of filtered radiographic images DATA2, the implementing means 8, the training means 10, and the comparing means 12.
  • the comparing means 12 deliver a data set DATA4 indicative of the classification accuracy of the deep-learning algorithm with respect to the classification performed by the statistical learning algorithm .
  • Analysis of the data set DATA4 allows the operation of the deep-learning algorithm ALGO2 to be validated.
  • new data sets DATA1 are generated with a view to continuing the training of the deep-learning algorithm ALGO2.
  • the system 1 is implemented on the basis of said set with a view to continuing the training of the deep-learning algorithm ALGO2 and improving the accuracy of said algorithm .
  • Figure 2 illustrates one example of implementation of the system 1 .
  • the capturing means 3 capture the set of two-dimensional digital radiographic images DATA1 of rolling elements 2 comprising defects and not comprising any defects.
  • the set of images DATA1 for example comprises 100 images, allowing a sufficient classification accuracy to be obtained from the algorithm ALGO1 .
  • the filtering means 6 filter the images of the set of images DATA1 , so as to improve the contrast of each image, and deliver the set of filtered images DATA2 to the generating means 7 and to the training means 10.
  • the generating means 7 generate the data set DATA3 on the basis of the set of filtered images DATA2 (step 24)
  • the implementing means 8 implement the statistical learning algorithm ALGO1 so that the statistical learning algorithm ALGO1 classifies each image of the set of filtered images DATA2 into the class CL1 of suspect rolling elements or into the class CL2 of non-suspect rolling elements (step 26) .
  • the training means 10 train the deep-learning algorithm ALGO2 so that the deep-learning algorithm classifies each image of the set of filtered images DATA2 into the class CL1 of suspect rolling elements or into the class CL2 of non-suspect rolling elements (step 28) .
  • the two steps 24 and 26, and step 28 are for example carried out in parallel.
  • steps 24 and 26, and 28 are carried out sequentially.
  • the generating means 7 segment each image of the set of filtered images DATA2 into various regions.
  • the means 7 For each region of each image of the set of filtered images DATA2, the means 7 generate a data vector containing the coding of each image region of the set of images DATA2.
  • the data set DATA3 comprises the data vector.
  • Figure 3 illustrates an image, filtered by the filtering means 6, of a rolling element 2.
  • Figure 4 illustrates division of the filtered image shown in figure 3 into three different regions Z1 , Z2, Z3, each region being characterized by a different greyscale level.
  • Each image is filtered by greyscale level.
  • each image may be divided into more than at least three different regions.
  • Figure 5 illustrates one example of a data set DATA3 modelled by coding lines Lc.
  • Each line Lc represents data coding an image of the set of filtered images DATA2.
  • the algorithm ALGO1 implemented by the implementing means 8 computes a value of an image- analysis indicator on the basis of the data set DATA3 for each image of the set of images DATA2.
  • the image-analysis indicator for example comprises computing an average, and/or a standard deviation, and/or a median, and/or a greyscale gradient.
  • the value of the image indicator is compared to a threshold.
  • the training means 10 segment each image of the set of filtered images DATA2 into a plurality of regions, each region corresponding to one selection criterion for example comprising the geometry of the rolling element 2 in said region, and/or the texture in said region and/or the contrast in said region.
  • the algorithm ALGO2 implemented by the training means 10 processes each region of each image to detect at least one defect on the basis of the selection criterion, detects any defects in the rolling element 2 and classifies said rolling element into one of the classes CL1 and CL2 depending on whether the rolling element 2 is considered to be suspect or not.
  • the algorithm ALGO2 classifies said rolling element 2 into the class CL2.
  • the comparing means 12 compare the classification performed by the statistical learning algorithm ALGO1 , which classification is stored in the memory 9 of the means 8, and the classification performed by the deep-learning algorithm ALGO2, which classification is stored in the memory 1 1 , for each given image of the set of images DATA1 .
  • the comparing means 12 determ ine the number of images classified into the same class CL1 , CL2 by the statistical learning algorithm ALGO1 and by the deep-learning algorithm ALGO2, and determ ine the classification accuracy of the deep-learning algorithm ALGO2 by computing the ratio between the number of images classified into the same class by both algorithms ALGO1 , ALGO2 and the total number of images of the set of radiographic images DATA1 .
  • the comparing means 12 deliver the data set DATA4.
  • the procedure for recognition of defects in the rolling element may be automated so as to increase the production rate of a line for producing rolling elements while m inim izing the risks of error in the interpretation of the image set DATA1 .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Mechanical Engineering (AREA)
  • Software Systems (AREA)
  • Medical Informatics (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)
  • Analysing Materials By The Use Of Radiation (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

This system (1) for training a deep-learning algorithm (ALGO2) for detecting a defect in a ceramic rolling element (2), comprises: - means (3) for capturing a set of digital radiographic images (DATA1) of rolling elements (2) comprising defects and not comprising any defects, - filtering means (6), - means (7) for generating a data set (DATA3) on the basis of the set of filtered radiographic images (DATA2), - implementing means (8), - training means (10) configured to train a deep-learning algorithm (ALGO2) on the basis of the set of radiographic images (DATA1), and - comparing means (12) configured to compare the classification of the images of the image set performed by the statistical learning algorithm (ALGO1) and the classification of the images of the image set performed by the deep-learning algorithm (ALGO2) with a view to determining the accuracy of the classification of the deep-learning algorithm with respect to the classification performed by the statistical learning algorithm.

Description

Description
Title: System for training a deep-learning algorithm and associated method
Technical field of the invention
The present invention relates to ceram ic rolling elements, and more particularly relates to the detection of manufacturing defects in such rolling elements.
Prior art
I n order to guide a mechanical assembly in rotation, it is generally proposed to use rolling bearings equipped with rolling elements made of ceram ic or of steel. The rolling elements may for example be balls or even cylindrical, conical or spherical rollers.
These rolling elements of rolling bearings allow a circular movement of a shaft with respect to a fixed element to be ensured while lim iting frictional force.
They are applicable to reversible electric motors or combustion engines, both in the automotive and aeronautical fields.
However, as ceramic rolling elements are produced by sintering, they may contain inclusions of foreign material, non-uniform agglomerations of material, or porosities.
These defects are liable to make the rolling elements fail and therefore to damage the product comprising the rolling bearing equipped with such rolling elements.
For example, when the rolling bearing equipped with failed rolling elements is located on a rotating shaft, they may cause spalling and/or overheating and the shaft will rotate with increasing difficulty.
Moreover, the rolling elements are liable to completely disunite from the other components of the bearing and thus cause the rotating shaft to disassociate from its surrounding mechanical system : this may prove to be critical in aeronautics.
To prevent failure of a rolling element, it is advantageous to detect these defects right after its manufacture.
It is known to use x-rays to obtain radiographic images of rolling elements. These images, after processing, are analysed by a human operator with a view to recognition of defects in the rolling element.
However, the procedure of recognition of defects in a rolling element by an operator is unsuitable for mass manufactured rolling elements.
As the ability of an operator to process images is lim ited, the defectrecognition procedure may slow the entirety of the rolling-element production line.
Furthermore, the operator may make m istakes in his interpretation of the images.
It is known to use a deep-learning algorithm , a neural network for example, to recognize predetermined features in an image.
However, in order for a deep-learning algorithm to perform well it must be trained.
I n light of the above, the goal of the invention is to overcome the aforementioned constraints.
Sum mary of the invention
One subject of the invention is therefore a method for training a deeplearning algorithm for detecting a defect in a ceram ic rolling element, comprising:
-capturing a set of digital radiographic images of rolling elements comprising defects and not comprising any defects,
-filtering a first time each image of the set of images in order to improve the contrast of each image,
-generating a data set on the basis of the set of filtered radiographic images,
-classifying, on the basis of the data set, each image of the set of radiographic images using a statistical learning algorithm into a class of suspect rolling elements or into a class of non-suspect rolling elements,
-training, on the basis of the set of radiographic images, a deep-learning algorithm so that the deep-learning algorithm classifies each image of the set of radiographic images into the class of suspect rolling elements or into the class of non-suspect rolling elements, and
-comparing the classification of the images of the image set performed by the statistical learning algorithm and the classification of the images of the image set performed by the deep-learning algorithm with a view to determ ining the accuracy of the classification of the deep-learning algorithm with respect to the classification performed by the statistical learning algorithm .
The analysis of the data of the classification performed by the learning algorithm makes it possible to interpret the variation in the weights of the deeplearning algorithm when it comprises a neural network.
When the precision of the deep-learning algorithm is sufficient, the procedure for recognition of defects in the rolling element may be automated so as to increase the production rate of a line for producing rolling elements while m inimizing the risks of error in the interpretation of the image set.
Preferably, the generation of a training data set comprises: -segmenting each image into various regions, -filtering a second time each region of each image of the set of images to obtain a data vector, the data vector forming the training data set.
Advantageously, the classification of each image of the set of radiographic images by a statistical learning algorithm comprises:
-computing a value of an image-analysis indicator on the basis of the indicative data set of each image of the set of filtered images,
-comparing the analysis indicator to a threshold, and
-classifying each image depending on the result of the comparison.
Preferably, the image-analysis indicator comprises computing an average, and/or a standard deviation, and/or a median, and/or a greyscale gradient.
Advantageously, the training of a deep-learning algorithm comprises: -segmenting each filtered image into various regions, and -processing each region of each image using the deep-learning algorithm with a view to detecting at least one defect on the basis of a selection criterion, -detecting a defect in light of a selection criterion specific to each region, and
-classifying each image.
Preferably, the selection criterion comprises the geometry of the rolling element in said region, and/or texture in said region and/or contrast in said region. Advantageously, the deep-learning algorithm comprises a convolutional neural network.
Preferably, the comparison of the classification of the images of the image set performed by the statistical learning algorithm and of the classification of the images of the image set performed by the deep-learning algorithm comprises:
-comparing the classification performed by the statistical learning algorithm and the classification performed by the deep-learning algorithm for each same image of the set of radiographic images,
-determ ining the number of images classified into the same class by the statistical learning algorithm and by the deep-learning algorithm , and
-determ ining the classification accuracy of the deep-learning algorithm by computing the ratio between the number of images classified into the same class by the two algorithms and the total number of images in the set of radiographic images.
Another subject of the invention is a system for training a deep-learning algorithm for detecting a defect in a ceram ic rolling element, comprising:
-means for capturing a set of digital radiographic images of rolling elements comprising defects and not comprising any defects,
-filtering means configured to improve the contrast of each image of the set of images,
-means for generating a data set on the basis of the set of filtered radiographic images,
-implementing means configured to implement a statistical learning algorithm configured to classify, on the basis of the data set, each image of the image set into a class of suspect rolling elements or into a class of non-suspect rolling elements,
-training means configured to train, on the basis of the set of radiographic images, a deep-learning algorithm so that the deep-learning algorithm classifies each image of the set of radiographic images into the class of suspect rolling elements or into the class of non-suspect rolling elements, and
-comparing means configured to compare the classification of the images of the image set performed by the statistical learning algorithm and the classification of the images of the image set performed by the deep-learning algorithm with a view to determ ining the accuracy of the classification of the deep-learning algorithm with respect to the classification performed by the statistical learning algorithm .
Brief description of the figures
Other aims, features and advantages of the invention will become apparent on reading the following description, which is given purely by way of non-lim iting example, and with reference to the appended drawings, in which:
[ Fig 1 ] illustrates a system for training a deep-learning algorithm for detecting a defect in a ceram ic rolling element, according to one embodiment of the invention,
[ Fig 2] shows one mode of implementation of the training system according to the invention,
[ Fig 3] shows one example of a filtered image according to the invention, [ Fig 4] shows one example of division of a filtered image, and
[ Fig 5] shows one example of a data set according to the invention. Detailed description of the invention
Figure 1 shows a system for training a deep-learning algorithm ALG1 for detecting a defect in a ceram ic rolling element 2.
The system 1 comprises means 3 for capturing a set of two-dimensional digital radiographic images DATA1 of rolling elements 2 comprising defects and not comprising any defects,
The capturing means 3 comprise a photographic sensor 4 and a generator 5 of high-frequency electromagnetic waves R1 configured to em it the electromagnetic waves R1 towards the element 2.
These electromagnetic waves R1 are subsequently absorbed by the photographic sensor 4.
A two-dimensional digital radiographic image of the rolling element 2 is thus formed.
It will be noted that the generator 4 generally takes the form of a m icrofocus x-ray tube.
According to another embodiment, the capturing means 3 generate a three-dimensional digital radiographic image.
The system 1 further comprises filtering means 6 that are connected to the capturing means 3 and that process the images of the set of images to improve the contrast of each image of the set of images DATA1 captured by the capturing means 3. The filtering means 6 deliver a set of filtered radiographic images DATA2.
The system 1 also comprises means 7 for generating a data set DATA3 from the set of radiographic images DATA2 filtered by the filtering means 6, and implementing means 8 that implement a statistical learning algorithm ALGO1 that is configured to classify, on the basis of the data set DATA3, each image of the set of filtered images DATA2 into a class CL1 of suspect rolling elements 2 or into a class CL2 of non-suspect rolling elements 2. The generating means 7 are connected to the filtering means 6 and to the implementing means 8.
The classification performed by the statistical learning algorithm ALGO1 is for example stored in a memory 9.
The statistical learning algorithm ALGO1 for example comprises a decision-tree algorithm , a random-forest algorithm , a support-vector-machine algorithm , a K-nearest-neighbours algorithm , or a logistic-regression algorithm .
It will be noted that a suspect rolling element is a rolling element 2 the probability of having at least one defect of which is higher than its probability of not having any defects.
Defects for example comprise inclusions of foreign material, non-uniform agglomerations of material or porosities.
The system 1 further comprises training means 10 that train, on the basis of the set of filtered radiographic images DATA2, a deep-learning algorithm ALGO2 so that the deep-learning algorithm classifies each image of the set of filtered radiographic images DATA2 into the class CL1 of suspect rolling elements or into the class CL2 of non-suspect rolling elements. The training means 10 are connected to the filtering means 6.
The deep-learning algorithm ALGO2 for example comprises a convolutional neural network.
The classification performed by the deep-learning algorithm ALGO2 is for example stored in a memory 1 1 .
The system 1 further comprises comparing means 12 that compare the classification performed by the statistical learning algorithm ALGO1 , and the classification performed by the deep-learning algorithm ALGO2 in order to determ ine the accuracy of the deep-learning algorithm on the basis of the classification of the statistical learning algorithm . The comparing means 12 are connected to the implementing means 8 and to the training means 10, and for example read the content of the memory 9 of the implementing means storing the classification performed by the statistical learning algorithm ALGO1 and the content of the memory 1 1 of the training means 10 storing the classification performed by the deep-learning algorithm ALGO2.
The system 1 further comprises a processing unit UT that employs the capturing means 3 to capture the set of images DATA1 , the filtering means, means 7 for generating a data set DATA3 on the basis of the set of filtered radiographic images DATA2, the implementing means 8, the training means 10, and the comparing means 12.
The comparing means 12 deliver a data set DATA4 indicative of the classification accuracy of the deep-learning algorithm with respect to the classification performed by the statistical learning algorithm .
Analysis of the data set DATA4 allows the operation of the deep-learning algorithm ALGO2 to be validated.
If the accuracy level is insufficient, new data sets DATA1 are generated with a view to continuing the training of the deep-learning algorithm ALGO2.
As a variant, when a set of images DATA1 of sufficient size is available, the system 1 is implemented on the basis of said set with a view to continuing the training of the deep-learning algorithm ALGO2 and improving the accuracy of said algorithm .
Figure 2 illustrates one example of implementation of the system 1 .
I n a step 20, the capturing means 3 capture the set of two-dimensional digital radiographic images DATA1 of rolling elements 2 comprising defects and not comprising any defects.
The set of images DATA1 for example comprises 100 images, allowing a sufficient classification accuracy to be obtained from the algorithm ALGO1 .
When the set of the images DATA1 is complete, in a step 22, the filtering means 6 filter the images of the set of images DATA1 , so as to improve the contrast of each image, and deliver the set of filtered images DATA2 to the generating means 7 and to the training means 10. On receipt of the set of filtered images DATA2, the generating means 7 generate the data set DATA3 on the basis of the set of filtered images DATA2 (step 24) , then the implementing means 8 implement the statistical learning algorithm ALGO1 so that the statistical learning algorithm ALGO1 classifies each image of the set of filtered images DATA2 into the class CL1 of suspect rolling elements or into the class CL2 of non-suspect rolling elements (step 26) .
Furthermore, on receipt of the set of filtered images DATA2, the training means 10 train the deep-learning algorithm ALGO2 so that the deep-learning algorithm classifies each image of the set of filtered images DATA2 into the class CL1 of suspect rolling elements or into the class CL2 of non-suspect rolling elements (step 28) .
The two steps 24 and 26, and step 28 are for example carried out in parallel.
As a variant, steps 24 and 26, and 28 are carried out sequentially.
I n step 24 of generating the data set DATA3, the generating means 7 segment each image of the set of filtered images DATA2 into various regions.
Next, for each region of each image of the set of filtered images DATA2, the means 7 generate a data vector containing the coding of each image region of the set of images DATA2.
The data set DATA3 comprises the data vector.
Figure 3 illustrates an image, filtered by the filtering means 6, of a rolling element 2.
Figure 4 illustrates division of the filtered image shown in figure 3 into three different regions Z1 , Z2, Z3, each region being characterized by a different greyscale level.
Each image is filtered by greyscale level.
Of course, each image may be divided into more than at least three different regions.
Figure 5 illustrates one example of a data set DATA3 modelled by coding lines Lc.
Each line Lc represents data coding an image of the set of filtered images DATA2.
When the implementing means 8 receive the data set DATA3, in step 26, the algorithm ALGO1 implemented by the implementing means 8 computes a value of an image- analysis indicator on the basis of the data set DATA3 for each image of the set of images DATA2.
The image-analysis indicator for example comprises computing an average, and/or a standard deviation, and/or a median, and/or a greyscale gradient.
Next, the value of the image indicator is compared to a threshold.
If the value of the indicator of an image of the set of images DATA2 is higher than the threshold, said image is for example classified into the class of suspect rolling elements CL1 , and if the value of the indicator of an image is lower than the threshold, said image is for example classified into the class of non-suspect rolling elements CL2.
I n step 28, the training means 10 segment each image of the set of filtered images DATA2 into a plurality of regions, each region corresponding to one selection criterion for example comprising the geometry of the rolling element 2 in said region, and/or the texture in said region and/or the contrast in said region.
Next, the algorithm ALGO2 implemented by the training means 10 processes each region of each image to detect at least one defect on the basis of the selection criterion, detects any defects in the rolling element 2 and classifies said rolling element into one of the classes CL1 and CL2 depending on whether the rolling element 2 is considered to be suspect or not.
For example, if the geometry of the element 2 is not circular to within a predefined tolerance in the region relative to the geometry of the element 2, the algorithm ALGO2 classifies said rolling element 2 into the class CL2.
When all the images of the set of filtered images DATA2 have been classified into one of the classes CL1 and CL2 by the algorithms ALGO1 and ALGO2, in step 30, the comparing means 12 compare the classification performed by the statistical learning algorithm ALGO1 , which classification is stored in the memory 9 of the means 8, and the classification performed by the deep-learning algorithm ALGO2, which classification is stored in the memory 1 1 , for each given image of the set of images DATA1 .
The comparing means 12 determ ine the number of images classified into the same class CL1 , CL2 by the statistical learning algorithm ALGO1 and by the deep-learning algorithm ALGO2, and determ ine the classification accuracy of the deep-learning algorithm ALGO2 by computing the ratio between the number of images classified into the same class by both algorithms ALGO1 , ALGO2 and the total number of images of the set of radiographic images DATA1 .
Next, the comparing means 12 deliver the data set DATA4. The analysis of the data DATA4 of the classification performed by the statistical learning algorithm ALGO1 , which classification is stored in the memory 9, makes it possible to interpret the variation in the weights of the deep-learning algorithm ALGO2 when it comprises a neural network.
When the precision of the deep-learning algorithm ALGO2 is sufficient, the procedure for recognition of defects in the rolling element may be automated so as to increase the production rate of a line for producing rolling elements while m inim izing the risks of error in the interpretation of the image set DATA1 .

Claims

Claims
1 . Method for training a deep-learning algorithm (ALGO2) for detecting a defect in a ceramic rolling element (2) , comprising: capturing a set of digital radiographic images (DATA1 ) of rolling elements (2) comprising defects and not comprising any defects, filtering a first time each image of the set of images (DATA1 ) in order to improve the contrast of each image, generating a data set (DATA3) on the basis of the set of filtered radiographic images (DATA2) , classifying, on the basis of the data set (DATA3) , each image of the set of radiographic images (DATA1 ) using a statistical learning algorithm (ALGO1 ) into a class (CL1 ) of suspect rolling elements or into a class (CL2) of non-suspect rolling elements, training, on the basis of the set of radiographic images (DATA1 ), a deep-learning algorithm (ALGO2) so that the deep-learning algorithm classifies each image of the set of radiographic images into the class (CL1 ) of suspect rolling elements or into the class (CL2) of non-suspect rolling elements, and comparing the classification of the images of the image set performed by the statistical learning algorithm (ALGO1 ) and the classification of the images of the image set performed by the deep-learning algorithm (ALGO2) with a view to determ ining the accuracy of the classification of the deeplearning algorithm with respect to the classification performed by the statistical learning algorithm .
2. Method according to Claim 1 , wherein the generation of a training data set (DATA3) comprises: segmenting each image into various regions (Z1 , Z2, Z3) , filtering a second time each region of each image of the set of images to obtain a data vector, the data vector forming the training data set (DATA3) .
3. Method according to either of Claims 1 and 2, wherein the classification of each image of the set of radiographic images by a statistical learning algorithm (ALGO1 ) comprises: computing a value of an image-analysis indicator on the basis of the indicative data set (DATA3) of each image of the set of filtered images (DATA2) , comparing the analysis indicator to a threshold, and classifying each image depending on the result of the comparison.
4. Method according to Claim 3, wherein the image-analysis indicator comprises computing an average, and/or a standard deviation, and/or a median, and/or a greyscale gradient.
5. Method according to any one of the preceding claims, wherein the training of a deep-learning algorithm (ALGO2) comprises: segmenting each filtered image into a plurality of regions (Z1 , Z2, Z3), processing each region of each image using the deep-learning algorithm (ALGO2) with a view to detecting at least one defect on the basis of a selection criterion, detecting a defect in light of a selection criterion specific to each region, and classifying each image.
6. Method according to Claim 5, wherein the selection criterion comprises the geometry of the rolling element in said region, and/or texture in said region and/or contrast in said region.
7. Method according to any one of Claims 1 to 4, wherein the deeplearning algorithm (ALGO2) comprises a convolutional neural network.
8. Method according to any one of the preceding claims, wherein the comparison of the classification of the images of the image set performed by the statistical learning algorithm (ALGO1 ) and of the classification of the images of the image set performed by the deep-learning algorithm (ALGO2) comprises: comparing the classification performed by the statistical learning algorithm (ALGO1 ) and the classification performed by the deep-learning algorithm (ALGO2) for each same image of the set of radiographic images, determ ining the number of images classified into the same class (CL1 , CL2) by the statistical learning algorithm and by the deep-learning algorithm , and determ ining the classification accuracy of the deep-learning algorithm by computing the ratio between the number of images classified into the same class by the two algorithms and the total number of images in the set of radiographic images.
9. System (1 ) for training a deep-learning algorithm (ALGO2) for detecting a defect in a ceram ic rolling element (2), comprising: means (3) for capturing a set of digital radiographic images (DATA1 ) of rolling elements (2) comprising defects and not comprising any defects, filtering means (6) configured to improve the contrast of each image of the set of images (DATA1 ), means (7) for generating a data set (DATA3) on the basis of the set of filtered radiographic images (DATA2) , implementing means (8) configured to implement a statistical learning algorithm (ALGO1 ) configured to classify, on the basis of the data set (DATA2) , each image of the image set (DATA1 ) into a class (CL1 ) of suspect rolling elements or into a class (CL2) of non-suspect rolling elements, training means ( 10) configured to train, on the basis of the set of radiographic images (DATA1 ), a deep-learning algorithm (ALGO2) so that the deep-learning algorithm classifies each image of the set of radiographic images into the class (CL1 ) of suspect rolling elements or into the class (CL2) of nonsuspect rolling elements, and comparing means ( 12) configured to compare the classification of the images of the image set performed by the statistical learning algorithm (ALGO1 ) and the classification of the images of the image set performed by the deeplearning algorithm (ALGO2) with a view to determ ining the accuracy of the classification of the deep-learning algorithm with respect to the classification performed by the statistical learning algorithm .
PCT/EP2022/079878 2021-11-08 2022-10-26 System for training a deep-learning algorithm and associated method Ceased WO2023078747A1 (en)

Priority Applications (7)

Application Number Priority Date Filing Date Title
CN202280074107.3A CN118302800A (en) 2021-11-08 2022-10-26 Systems and related methods for training deep learning algorithms
KR1020247018217A KR20240118084A (en) 2021-11-08 2022-10-26 Systems and associated methods for training deep learning algorithms
GB2405441.3A GB2625498A (en) 2021-11-08 2022-10-26 System for training a deep-learning algorithm and associated method
JP2024525899A JP2024543023A (en) 2021-11-08 2022-10-26 SYSTEM FOR TRAINING DEEP LEARNING ALGORITHMS AND RELATED METHODS
CA3236753A CA3236753A1 (en) 2021-11-08 2022-10-26 System for training a deep-learning algorithm and associated method
US18/706,495 US20250005734A1 (en) 2021-11-08 2022-10-26 System for training a deep-learning algorithm and associated method
DE112022005358.5T DE112022005358T5 (en) 2021-11-08 2022-10-26 System for training a deep learning algorithm and associated method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FRFR2111796 2021-11-08
FR2111796A FR3129020B1 (en) 2021-11-08 2021-11-08 System for training a deep learning algorithm and associated method

Publications (1)

Publication Number Publication Date
WO2023078747A1 true WO2023078747A1 (en) 2023-05-11

Family

ID=79601734

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2022/079878 Ceased WO2023078747A1 (en) 2021-11-08 2022-10-26 System for training a deep-learning algorithm and associated method

Country Status (9)

Country Link
US (1) US20250005734A1 (en)
JP (1) JP2024543023A (en)
KR (1) KR20240118084A (en)
CN (1) CN118302800A (en)
CA (1) CA3236753A1 (en)
DE (1) DE112022005358T5 (en)
FR (1) FR3129020B1 (en)
GB (1) GB2625498A (en)
WO (1) WO2023078747A1 (en)

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR3102554B1 (en) * 2019-10-23 2021-11-19 Alstom Transp Tech Method and system for estimating the wear of a rotating machine comprising a bearing

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
CHOUDHARY ANURAG ET AL: "Convolutional neural network based bearing fault diagnosis of rotating machine using thermal images", MEASUREMENT, INSTITUTE OF MEASUREMENT AND CONTROL. LONDON, GB, vol. 176, 20 February 2021 (2021-02-20), XP086541378, ISSN: 0263-2241, [retrieved on 20210220], DOI: 10.1016/J.MEASUREMENT.2021.109196 *
DOMINGO MERY ET AL: "Automatic Defect Recognition in X-Ray Testing Using Computer Vision", CORR (ARXIV), 10 January 2017 (2017-01-10), pages 1026 - 1035, XP055405098, ISBN: 978-1-5090-4822-9, DOI: 10.1109/WACV.2017.119 *
HU CHUANFEI ET AL: "An Efficient Convolutional Neural Network Model Based on Object-Level Attention Mechanism for Casting Defect Detection on Radiography Images", vol. 67, no. 12, 1 December 2020 (2020-12-01), USA, pages 10922 - 10930, XP055930146, ISSN: 0278-0046, Retrieved from the Internet <URL:https://ieeexplore.ieee.org/stampPDF/getPDF.jsp?tp=&arnumber=8948332&ref=aHR0cHM6Ly9pZWVleHBsb3JlLmllZWUub3JnL2RvY3VtZW50Lzg5NDgzMzI=> [retrieved on 20220612], DOI: 10.1109/TIE.2019.2962437 *
JANSSENS OLIVIER ET AL: "Convolutional Neural Network Based Fault Detection for Rotating Machinery", JOURNAL OF SOUND AND VIBRATION, ELSEVIER, AMSTERDAM , NL, vol. 377, 24 May 2016 (2016-05-24), pages 331 - 345, XP029564152, ISSN: 0022-460X, DOI: 10.1016/J.JSV.2016.05.027 *
JIANG LILI ET AL: "Casting defect detection in X-ray images using convolutional neural networks and attention-guided data augmentation", MEASUREMENT, INSTITUTE OF MEASUREMENT AND CONTROL. LONDON, GB, vol. 170, 18 November 2020 (2020-11-18), XP086438678, ISSN: 0263-2241, [retrieved on 20201118], DOI: 10.1016/J.MEASUREMENT.2020.108736 *

Also Published As

Publication number Publication date
JP2024543023A (en) 2024-11-19
US20250005734A1 (en) 2025-01-02
DE112022005358T5 (en) 2024-08-22
FR3129020B1 (en) 2024-03-15
CN118302800A (en) 2024-07-05
CA3236753A1 (en) 2023-05-11
KR20240118084A (en) 2024-08-02
FR3129020A1 (en) 2023-05-12
GB2625498A (en) 2024-06-19
GB202405441D0 (en) 2024-06-05

Similar Documents

Publication Publication Date Title
CN108444708B (en) Method for establishing rolling bearing intelligent diagnosis model based on convolutional neural network
US11295173B2 (en) Image identification apparatus, image identification method, and recording medium
Ansari et al. A Convolutional Neural Network (CNN) classification to identify the presence of pores in powder bed fusion images
Chen et al. Dynamic vision enabled contactless cross-domain machine fault diagnosis with neuromorphic computing
CN110276416A (en) A kind of rolling bearing fault prediction technique
Kähler et al. Anomaly detection for industrial surface inspection: application in maintenance of aircraft components
CN103499585A (en) Non-continuity lithium battery thin film defect detection method and device based on machine vision
CN116754567B (en) Periodic defect detection method, device and equipment for copper foil material
Saha et al. Enhancing bearing fault diagnosis using transfer learning and random forest classification: A comparative study on variable working conditions
Huo et al. VDDNet: An object detection model for measuring the vibration displacement of rotating structure
Sato et al. Anomaly detection of laser powder bed fusion melt pool images using combined unsupervised and supervised learning methods
Kumar et al. A deep learning-based CNN-GRU prediction model for early fault diagnosis in rolling ball bearings
US20250005734A1 (en) System for training a deep-learning algorithm and associated method
Li et al. Fault diagnosis of the harmonic reducer based on dual-path convolutional network with multi-channel hybrid attention mechanism
Manoj et al. Porosity prediction of 3d printed components using u-net and its variants
CN112861675B (en) Detection and identification method for fecal component
Tao et al. Enhancing keyhole porosity detection in laser powder bed fusion using a hybrid approach of machine learning and physics-informed knowledge
Yan et al. Transforming PCB Solder Joint Detection with Deep Learning Empowered X-ray Nondestructive Testing
Im et al. Application of image classification model to improve visual inspection systems of industrial cutters
CN117808815B (en) Method and device for detecting quality of outer wall of implantation and intervention tubular instrument
Mir et al. Proactive Management of Shrinkage Defects using Deep Learning Analytics
Banerjee et al. Manufacturing Flaw Forecasting via CNN-RF Fusion Techniques
Najim et al. Detection of roller bearing faults at different rotational speeds using envelope analysis features and support vector machine
Vishnupriyan et al. AI-Based Fault Detection in Power Electronics using Vibration/Current Signals
Li et al. Residual Dual Encoder Network using Distance Metric Learning for Intelligent Fault Recognition with Unknown Classes

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: 22805890

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2405441.3

Country of ref document: GB

WWE Wipo information: entry into national phase

Ref document number: 3236753

Country of ref document: CA

ENP Entry into the national phase

Ref document number: 2024525899

Country of ref document: JP

Kind code of ref document: A

WWE Wipo information: entry into national phase

Ref document number: 18706495

Country of ref document: US

WWE Wipo information: entry into national phase

Ref document number: 112024000079438

Country of ref document: IT

WWE Wipo information: entry into national phase

Ref document number: 202280074107.3

Country of ref document: CN

WWE Wipo information: entry into national phase

Ref document number: 112022005358

Country of ref document: DE

WWP Wipo information: published in national office

Ref document number: 2405441.3

Country of ref document: GB

122 Ep: pct application non-entry in european phase

Ref document number: 22805890

Country of ref document: EP

Kind code of ref document: A1

WWG Wipo information: grant in national office

Ref document number: 112024000079438

Country of ref document: IT