CN113012123A - Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material - Google Patents

Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material Download PDF

Info

Publication number
CN113012123A
CN113012123A CN202110271298.XA CN202110271298A CN113012123A CN 113012123 A CN113012123 A CN 113012123A CN 202110271298 A CN202110271298 A CN 202110271298A CN 113012123 A CN113012123 A CN 113012123A
Authority
CN
China
Prior art keywords
defect
carbon fiber
composite material
fiber composite
quantitative analysis
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.)
Pending
Application number
CN202110271298.XA
Other languages
Chinese (zh)
Inventor
姜明顺
丁国强
贾磊
曹弘毅
张雷
张法业
隋青美
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.)
Shandong University
Original Assignee
Shandong University
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 Shandong University filed Critical Shandong University
Priority to CN202110271298.XA priority Critical patent/CN113012123A/en
Publication of CN113012123A publication Critical patent/CN113012123A/en
Pending legal-status Critical Current

Links

Images

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
    • G06T7/0008—Industrial image inspection checking presence/absence
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2431—Multiple classes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—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
    • 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
    • G06T2207/10—Image acquisition modality
    • G06T2207/10132—Ultrasound image
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Multimedia (AREA)
  • Geometry (AREA)
  • Quality & Reliability (AREA)
  • Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)

Abstract

The invention discloses a method and a system for classifying, identifying and quantitatively analyzing defect and damage of a carbon fiber composite material, wherein the method comprises the following steps: constructing a defect data set of the carbon fiber composite material; training the constructed convolutional neural network model with residual errors by using the data set; acquiring real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, determining relative position information of the defect, and performing classification identification and quantitative analysis on the defect image by using a trained convolutional neural network. The invention has the beneficial effects that: the method has the advantages that the method can automatically extract the characteristics of different defects to carry out real-time online classification recognition and quantitative analysis, so that the problem of low classification precision caused by inaccurate defect characteristics extracted manually is solved, the capacity of batch defect classification recognition and quantitative analysis is improved, and a technical basis is especially provided for carrying out full-automatic defect detection and quantitative analysis of the carbon fiber composite material.

Description

Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material
Technical Field
The invention relates to the technical field of nondestructive testing, in particular to a method and a system for classifying, identifying and quantitatively analyzing defects and damages of a carbon fiber composite material.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The carbon fiber composite material is a novel material mainly formed by compounding high-performance carbon fibers with a matrix such as resin, metal, ceramic and the like, has the advantages of high specific modulus, high specific strength, light weight, fatigue resistance and the like, and is widely used in the fields of aerospace, rail transit, automobile manufacturing and the like.
Due to the particularity and complexity of the manufacturing process, carbon fiber composite materials inevitably bring about manufacturing defects in the production and manufacturing stage. During long-term use, various external environmental challenges are faced at any time, and even simple foreign object impact can cause damage to internal structures. Therefore, the inspection and monitoring of the internal structure of the carbon fiber composite material are continuously researched and developed.
In the field of nondestructive testing at the present stage, the carbon fiber composite material mainly adopts an ultrasonic testing mode to emit ultrasonic pulses with a certain frequency to an object to be tested, and generates graphs such as A-scan, B-scan and C-scan of the object to be tested according to received ultrasonic echo signals, so as to judge the structural health condition in the material. In the aspect of qualitative judgment and quantitative analysis of defects, professional detection personnel are required to analyze and judge the ultrasonic detection images with the defects one by one through professional analysis software matched with an ultrasonic detection system so as to determine the types of the defects, the sizes of the defects and the like, the process depends heavily on experience knowledge of the professional detection personnel, and not only is the workload large and the timeliness low, but also a certain misjudgment probability exists.
Disclosure of Invention
In order to solve the problems, the invention provides a method and a system for classifying, identifying and quantitatively analyzing defects and damages of a carbon fiber composite material. The system is based on a gray image generated by an ultrasonic C scanning signal (namely the maximum value of the amplitude of the ultrasonic A scanning signal of each coordinate point position in a set gate range), different target features are accurately extracted by adopting a CNN classification method, the relative position of the defect is automatically determined by an image processing technology, and finally real-time online classification recognition and quantitative analysis of the impact defect and the layered defect are realized, so that a technical basis is provided for automatic detection and judgment of the defect.
In some embodiments, the following technical scheme is adopted:
a classification, identification and quantitative analysis method for defect and damage of a carbon fiber composite material comprises the following steps:
constructing a defect data set of the carbon fiber composite material;
training the constructed convolutional neural network model with residual errors by using the data set;
acquiring real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, determining relative position information of the defect, and performing classification identification and quantitative analysis on the defect image by using a trained convolutional neural network.
As a further scheme, the process of constructing the defect data set of the carbon fiber composite material specifically includes:
manufacturing a carbon fiber composite material sample plate containing layered defects of different depths and different sizes; scanning the defect sample plate by using an ultrasonic scanning system, and performing ultrasonic detection by using a pulse-echo mode to obtain ultrasonic detection C scanning data of a defect-free region and layered defect regions with different depths and sizes; the classification standard of the layered defect is described by using a defect diameter dia according to practical conditions such as the detection resolution of an ultrasonic detection system, and the following (unit mm) is defined:
1 < dia < 3: micro-layering (type D1);
dia is more than 3 and less than or equal to 6: minor delamination (type D2);
dia is more than 6 and less than or equal to 9: medium stratification (type D3);
dia is more than 9 and less than or equal to 12: larger stratification (type D4);
manufacturing a carbon fiber composite material sample plate containing impact defects with different energy magnitudes; scanning the defect sample plate by using an ultrasonic scanning system, and performing ultrasonic detection by using a pulse-echo mode to obtain ultrasonic detection C scanning data of a defect-free area and impact defect areas with different energy sizes; the classification of the impact defect is divided into intervals according to the impact energy, and is defined as follows (unit J):
0 < energy is less than or equal to 10: low energy impact (type I1);
energy is more than 10 and less than or equal to 20: impact of medium energy (I2 type)
20 < energy < 30: high energy impact (I3 type)
Carrying out batch processing on the extracted ultrasonic detection C-scan data to obtain sample gray level pictures of different categories; and constructing a defect data set of the carbon fiber composite material.
As a further scheme, after obtaining sample grayscale pictures of different categories, the method further includes:
firstly, turning and then rotating, then changing the contrast brightness of an image, and finally adding random noise to perform data enhancement operation to amplify a detection sample; manufacturing a sample set to be detected according to the corresponding category label;
randomly disordering the sample set to be detected according to a preset proportion, and generating a training set, a verification set and a test set corresponding to the corresponding defect types.
As a further scheme, the convolutional neural network model with the residual error adopts the exponentially decaying dynamic learning rate to accelerate the training speed of the network model.
As a further scheme, the convolutional neural network model with residual errors is regularly detected through a verification set, and finally the optimal model hyper-parameters and weight values of each layer are obtained.
As a further scheme, the convolutional neural network model with the residual error automatically qualitatively divides a test set picture into a defect-free picture, a layering defect and an impact defect, quantitatively divides the layering defect into a D1 type layering defect, a D2 type layering defect, a D3 type layering defect and a D4 type layering defect, quantitatively divides the impact defect into an I1 type impact defect, an I2 type impact defect and an I3 type impact defect, and completes the classification identification and quantitative analysis of the impact defect and the layering defect of the carbon fiber composite material.
As a further scheme, real-time online ultrasonic detection C-scan data of a carbon fiber composite material to be detected is obtained by an ultrasonic phased array detection system, an original gray image is generated, whether the image has defects or not is judged by calculating the ratio of the average values of pixels in a set neighborhood where the maximum value and the minimum value of the image pixel are located according to the half-wave height method of ultrasonic detection, the threshold value of the image pixel is determined for the image with the defects, then binarization processing, opening operation and closing operation are carried out on the image, the defect area and the defect-free area are displayed in different colors, further outline information of the defect part is extracted, and the relative position of the defect area is determined.
As a further scheme, the original data image corresponding to the defect area is selected to meet the specification of the model requirement, and is input into the trained convolutional neural network model, and finally classification recognition and quantitative analysis of the defect of the carbon fiber composite material to be detected are achieved.
In other embodiments, the following technical solutions are adopted:
a carbon fiber composite material defect damage classification identification and quantitative analysis system comprises:
the data set construction module is used for constructing a defect data set of the carbon fiber composite material;
the neural network training module is used for training the constructed convolutional neural network model with the residual error by utilizing the data set;
and the defect positioning and identifying module is used for acquiring real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, determining relative position information of the defect, and performing classification identification and quantitative analysis on the defect image by using the trained convolutional neural network.
In other embodiments, the following technical solutions are adopted:
a terminal device comprising a processor and a memory, the processor being arranged to implement instructions; the memory is used for storing a plurality of instructions, and the instructions are suitable for being loaded by the processor and executing the carbon fiber composite material defect damage classification identification and quantitative analysis method.
In other embodiments, the following technical solutions are adopted:
a computer readable storage medium, wherein a plurality of instructions are stored, and the instructions are suitable for being loaded by a processor of a terminal device and executing the carbon fiber composite material defect damage classification identification and quantitative analysis method.
Compared with the prior art, the invention has the beneficial effects that:
the invention adopts a machine learning mode to automatically extract the characteristics of different defects, calibrate the relative positions of defect areas and perform classification identification and quantitative analysis of the defects on line in real time, thereby avoiding the problem of low classification precision caused by inaccurate manually extracted defect characteristics, improving the capacity of classification identification and quantitative analysis of the defects in batch processing, and particularly providing a technical basis for developing full-automatic defect detection and quantitative analysis of the carbon fiber composite material.
Additional features and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a schematic diagram of a convolutional neural network model training process with residual error according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a carbon fiber composite material artificial pre-buried layered defect sample plate in the embodiment of the invention;
FIG. 3 is a schematic diagram of a carbon fiber composite sample plate after a drop weight impact test is performed in an embodiment of the invention;
fig. 4 is a flow chart of real-time online detection of the carbon fiber composite material plate to be detected in the embodiment of the invention.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
Example one
The problem of classification, identification and quantitative analysis of defect and damage of the carbon fiber composite material essentially belongs to the problem of image classification. Image classification is an image processing method for distinguishing different types of targets according to different features in respective images, and is mainly characterized by extracting features of images. The convolutional neural network is a deep feedforward neural network containing convolutional calculation, is one of representative algorithms for deep learning, and compared with other algorithms, the convolutional neural network can automatically extract image features, so that the problem of low classification accuracy caused by improper artificial feature selection is solved. Therefore, the ultrasonic detection data of the carbon fiber composite material is combined with the CNN convolution network, and the automation and intelligence level of defect judgment of the carbon fiber composite material is improved on the premise of ensuring the reliable detection accuracy.
In view of the above, in one or more embodiments, a method for classifying, identifying and quantitatively analyzing defect damage of a carbon fiber composite material is disclosed, and with reference to fig. 1 and 4, the method includes the following steps:
constructing a defect data set of the carbon fiber composite material;
training the constructed convolutional neural network model with residual errors by using the data set;
acquiring real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, determining relative position information of the defect, and performing classification identification and quantitative analysis on the defect image by using a trained convolutional neural network.
The specific implementation steps are as follows:
(1) manufacturing a carbon fiber composite material sample plate containing layered defects of different depths and different sizes; scanning the defect sample plate by using an ultrasonic scanning system, and performing ultrasonic detection by using a pulse-echo mode to obtain ultrasonic detection C scanning data of a defect-free region and defect regions with different depths and sizes; the classification standard of the layered defect is described by using a defect diameter dia according to practical conditions such as the detection resolution of an ultrasonic detection system, and the following (unit mm) is defined:
1 < dia < 3: micro-layering (type D1);
dia is more than 3 and less than or equal to 6: minor delamination (type D2);
dia is more than 6 and less than or equal to 9: medium stratification (type D3);
dia is more than 9 and less than or equal to 12: larger stratification (type D4);
scanning the artificially pre-embedded layered defect carbon fiber composite sample plate by adopting an ultrasonic scanning system to obtain ultrasonic detection C scanning data of defects at different positions and different sizes;
the artificial pre-buried layered defect sample plate of the carbon fiber composite material is shown in figure 2, the overall size of the sample plate is 250 mm-200 mm, the thickness of the rectangular panel is 3mm, layered defects are prepared by pre-burying a polyimide film between layers, the layered defects are different in depth and size, the total number of the layered defects is 12, the layered defects are divided into four types of D1, D2, D3 and D4, the depth distance between adjacent layers is about 1mm, and the horizontal distance between adjacent defects is 50 mm. Specifically, the following table 1 shows.
TABLE 1 specification of layered Defect types
Figure BDA0002974536590000061
Figure BDA0002974536590000071
(2) Manufacturing a carbon fiber composite material sample plate containing impact defects with different energy magnitudes; scanning the defect sample plate by using an ultrasonic scanning system, and performing ultrasonic detection by using a pulse-echo mode to obtain ultrasonic detection C scanning data of a defect-free area and impact defect areas with different energy sizes; the classification of the impact defect is divided into intervals according to the impact energy, and is defined as follows (unit J):
0 < energy is less than or equal to 10: low energy impact (type I1);
energy is more than 10 and less than or equal to 20: impact of medium energy (I2 type)
20 < energy < 30: high energy impact (I3 type)
Scanning the carbon fiber composite material sample plate subjected to the drop hammer impact test by using an ultrasonic scanning system to obtain ultrasonic detection C scanning data of defects with different energy sizes;
the carbon fiber composite material plate is a sample plate with impact defect damage after a drop hammer impact test, the raw material is the carbon fiber composite material plate which is consistent with the layered defect sample plate and has no defect, the specification of the carbon fiber composite material plate is shown in figure 3, the rectangular panel with the overall size of 150mm x 100mm has no defect before the test, three sample plates with the same specification are respectively impacted by different impact energy in the test, and the impact center is positioned near the geometric center of the sample plate. The impact test yielded three specification (I1, I2, I3) impact defect panels representing low, medium and high energy impact defects, respectively. Specifically, as shown in table 2 below.
TABLE 2 impact Defect type Specification
Name (R) Type of impact defect Impact location
Template one I1 Geometric center
Sample plate two I2 Geometric center
Sample plate three I3 Geometric center
This example sample data acquisition all adopts ultrasonic phased array detecting system, and its host computer is OmniScan MX2 portable ultrasonic phased array defectoscope, and the phased array probe selects near-wall probe of linear array (5L64-NM1), and probe center frequency is 5MHZ, and the wafer interval is 1mm, and the wafer number is 64, and the organic glass voussoir that highly is 20mm is selected for use to the voussoir. Scanning the templates with different defect types in the steps (1) and (2) to obtain C scanning data of the defects with different types.
The C-scan data of the sample acquired in this example is the maximum value of the ultrasonic reflection echo signal (i.e., the amplitude of the a-scan signal) of each coordinate point of the sample in the gate range set by the ultrasonic phased array detection system. The data thus composed have the following advantages:
a. the data dimension is reduced, and the acquisition speed of the required sample data is improved;
b. the information of different types of defects and sizes of the defects can be stored on the basis of extremely small data volume;
c. the data processing difficulty is reduced, the training of the model can be accelerated, and the purpose of online real-time detection can be achieved;
(3) processing the acquired C-scan data to obtain different types of gray level pictures, then performing data enhancement operation to manufacture a sample set to be detected, randomly disordering the sample set according to a set proportion, and generating a training set, a verification set and a test set corresponding to corresponding defect type labels;
and (3) carrying out program processing on the original data acquired in the step (2), extracting data in batches, and adding corresponding labels to different types of defects so as to generate original sample pictures of different types, wherein the pictures adopt gray level graphs with 57 × 57 resolution and a single channel. And then performing data enhancement processing on the generated gray-scale images of different defect types, wherein the processing method can adopt image inversion, rotation, random noise addition, contrast brightness change and the like. In the embodiment, the original sample picture is turned over firstly and then rotated, then the contrast brightness of the image is changed, and finally random noise is added to carry out data enhancement operation. Wherein:
a. the image turning adopts three modes of horizontal turning, vertical turning and horizontal and vertical turning;
b. the image rotation adopts three angles of 90 degrees, 180 degrees and 270 degrees of anticlockwise rotation around the geometric center of the original image;
c. the formula adopted for changing the contrast brightness of the image is g (i, j) ═ α × f (i, j) + β, where g (i, j) is the value of the output image pixel, f (i, j) is the value of the input image pixel, α >0 is a gain parameter, the variation of the contrast is controlled, β is a bias parameter, and the variation of the brightness is controlled, and in this example, the parameters are used as (α ═ 1, β ═ 30) and (α ═ 1.5, β ═ 0);
d. the random noise is added by adding a noise point having a pixel value of 255 to 100 random coordinate positions in the picture.
Through data enhancement processing, different types of sample pictures with large data volume can be obtained and made into a sample set to be detected, then the sample set to be detected is subjected to data preprocessing, the sample sequence is disordered according to the classification labels according to the proportion of 6:2:2, and a training set, a verification set and a test set are made.
(4) Constructing a CNN network model with residual errors, training an optimization model by using data of a training set and a verification set, and storing the trained model, related hyper-parameters and weights to obtain a CNN classification model;
and (4) on the basis of the step (3), setting a model hyper-parameter for the CNN model constructed by inputting the training set data obtained in the step (3) to train. The CNN model architecture adopted in this example is based on the resnet convolutional neural network, and since the pixels of the picture to be trained are 57 × 57, the parameters padding of the first convolutional layer and the pooling layer of the convolutional neural network are set to be "same", and the parameters of the rest layers are unchanged. The optimizer of the CNN model adopts an exponentially decaying dynamic learning rate, and the parameters are set to (initial _ learning _ rate is 0.1, decay _ steps is 100, and decay _ rate is 0.96), so that the training speed of the network model can be increased. And the CNN model prevents the fitting condition from occurring through the regular detection of the verification set, finally obtains the appropriate model hyper-parameters and the weight values of all layers, and stores the model.
(5) And (3) putting the test set data into a trained CNN classification model, carrying out classification recognition and quantitative analysis on the defects according to the characteristics of the data by the model, and outputting a classification result.
And (4) inputting the test set pictures obtained in the step (3) into the CNN model obtained in the step (4), automatically qualitatively dividing the pictures into defects without defects, layered defects and impact defects, quantitatively dividing the layered defects into D1 type layered defects, D2 type layered defects, D3 type layered defects and D4 type layered defects, quantitatively dividing the impact defects into I1 type impact defects, I2 type impact defects and I3 type impact defects, and completing classification, identification and quantitative analysis of the impact defects and the layered defects of the carbon fiber composite material.
(6) Acquiring ultrasonic detection C scanning data of a carbon fiber composite material plate to be detected (the material is the same as that of a defect acquisition sample plate), generating an original gray picture of the material plate to be detected, judging whether the image has defects or not by calculating the ratio of the maximum value point to the minimum value point of the image pixel to the average value of the pixels in a 2 x2 neighborhood according to a half-wave height method of ultrasonic detection, determining the threshold value of the image pixel of the image containing the defects, carrying out binarization processing on the image, displaying black in the defect area and white in the non-defect area, further eliminating noise points in the image through opening operation and closing operation of the image, smoothing the image, then carrying out outline extraction on the defect part, and finally calibrating the relative position information of each defect in the original image.
In this example, the maximum value a of the original gray scale image pixels of the carbon fiber composite material plate to be detected is determined firstmaxAnd a minimum value AminRespectively selecting the pixel values in 2-by-2 neighborhood to calculate the mean value, and recording as
Figure BDA0002974536590000101
And
Figure BDA0002974536590000102
if it is
Figure BDA0002974536590000103
Then it is judged that it is defective, and then A is setmaxAnd/4, performing binarization processing on the image by taking the image as a threshold value to generate a binary image, adopting convolution kernels of (3, 3) to perform opening and closing operation on the image, finally extracting the outline of the defective part, and calibrating the relative position information of the defective part.
(7) And inputting the defect image data meeting the requirements of the convolutional neural network model into the trained model to obtain the classification category and energy information of the defect, and completing classification identification and quantitative analysis of the impact defect and the layering defect of the carbon fiber composite board to be detected.
And (4) on the basis of the step (6), acquiring image data meeting the specification required by the convolutional neural network model by taking the defect part marked by the original gray level image of the carbon fiber composite plate to be detected as the center, in the example, inputting an image with the defect pixels of 57 × 57 into the trained model, and finally completing the classification, identification and quantitative analysis of the impact defect and the layering defect of the carbon fiber composite plate to be detected.
Example two
In one or more embodiments, a system for classification, identification and quantitative analysis of defect damage of carbon fiber composite material is disclosed, comprising:
the data set construction module is used for constructing a defect data set of the carbon fiber composite material;
the neural network training module is used for training the constructed convolutional neural network model with the residual error by utilizing the data set;
and the defect positioning and identifying module is used for acquiring real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, determining relative position information of the defect, and performing classification identification and quantitative analysis on the defect image by using the trained convolutional neural network.
It should be noted that specific implementation manners of the modules are already described in the first embodiment, and are not described again.
EXAMPLE III
In one or more embodiments, a terminal device is disclosed, which includes a server, where the server includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the method for classifying and identifying and quantitatively analyzing defect damage of carbon fiber composite material in the first embodiment. For brevity, no further description is provided herein.
It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general purpose processors, digital signal processors DSP, application specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and so on. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may include both read-only memory and random access memory, and may provide instructions and data to the processor, and a portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.
In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or instructions in the form of software.
The method for classifying, identifying and quantitatively analyzing the defect and damage of the carbon fiber composite material in the first embodiment can be directly implemented by a hardware processor, or implemented by combining hardware and software modules in the processor. The software modules may be located in ram, flash, rom, prom, or eprom, registers, among other storage media as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and completes the steps of the method in combination with hardware of the processor. To avoid repetition, it is not described in detail here.
Those of ordinary skill in the art will appreciate that the various illustrative elements, i.e., algorithm steps, described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.

Claims (10)

1. A classification, identification and quantitative analysis method for defect and damage of a carbon fiber composite material is characterized by comprising the following steps:
constructing a defect data set of the carbon fiber composite material;
training the constructed convolutional neural network model with residual errors by using the data set;
acquiring real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, determining relative position information of the defect, and performing classification identification and quantitative analysis on the defect image by using a trained convolutional neural network.
2. The method for classification, identification and quantitative analysis of defect damage of carbon fiber composite material as claimed in claim 1, wherein the process of constructing the defect data set of carbon fiber composite material specifically comprises:
manufacturing a carbon fiber composite material sample plate containing layered defects of different depths and different sizes; scanning the defect sample plate by using an ultrasonic scanning system, and performing ultrasonic detection by using a pulse-echo mode to obtain ultrasonic detection C scanning data of a defect-free region and defect regions with different depths and sizes;
manufacturing a carbon fiber composite material sample plate containing impact defects with different energy magnitudes; scanning the defect sample plate by using an ultrasonic scanning system, and performing ultrasonic detection by using a pulse-echo mode to obtain ultrasonic detection C scanning data of a defect-free area and impact defect areas with different energy sizes;
carrying out batch processing on the extracted ultrasonic detection C-scan data to obtain sample gray level pictures of different categories; and constructing a defect data set of the carbon fiber composite material.
3. The method for classification, identification and quantitative analysis of defect and damage of carbon fiber composite material as claimed in claim 2, wherein after obtaining gray level images of samples of different classes, the method further comprises:
firstly, turning and then rotating, then changing the contrast brightness of an image, and finally adding random noise to perform data enhancement operation to amplify a detection sample; manufacturing a sample set to be detected according to the corresponding category label;
randomly disordering the sample set to be detected according to a preset proportion, and generating a training set, a verification set and a test set corresponding to the corresponding defect types.
4. The method for classification, identification and quantitative analysis of defect and damage of carbon fiber composite material as claimed in claim 1, wherein the convolutional neural network model with residual error adopts an exponentially decaying dynamic learning rate to accelerate the training speed of the network model.
5. The method for classification, identification and quantitative analysis of defect and damage of carbon fiber composite material as claimed in claim 1, wherein the convolutional neural network model with residual error is periodically detected through a verification set, and finally the best model hyper-parameter and weight value of each layer are obtained.
6. The method for classification, identification and quantitative analysis of defect damage of carbon fiber composite material as claimed in claim 1, wherein the convolutional neural network model with residual error automatically qualitatively classifies pictures into defect-free, layered defect and impact defect, then quantitatively classifies layered defect into D1 type layered defect, D2 type layered defect, D3 type layered defect and D4 type layered defect, and quantitatively classifies impact defect into I1 type impact defect, I2 type impact defect and I3 type impact defect, thereby completing classification, identification and quantitative analysis of impact defect and layered defect of carbon fiber composite material.
7. The method for classification, identification and quantitative analysis of defect damage of carbon fiber composite material as claimed in claim 1, wherein an ultrasonic phased array detection system is used to obtain real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, to generate an original gray image, according to a half-wave height method of ultrasonic detection, whether the image has a defect is judged by calculating a ratio between average values of pixels in a set neighborhood where points of a maximum value and a minimum value of the image pixel are located, a threshold value of the image pixel is determined for the image with the defect, then binarization processing, opening operation and closing operation are performed on the image, the defect area and the defect-free area are displayed in different colors, further contour information of the defect part is extracted, and the relative position of the defect area is determined;
and selecting the original data image corresponding to the defect area as a specification meeting the requirement of the model, inputting the specification into the trained convolutional neural network model, and finally realizing classification recognition and quantitative analysis of the defect of the carbon fiber composite material to be detected.
8. The utility model provides a carbon-fibre composite defect damage classification discernment and quantitative analysis system which characterized in that includes:
the data set construction module is used for constructing a defect data set of the carbon fiber composite material;
the neural network training module is used for training the constructed convolutional neural network model with the residual error by utilizing the data set;
and the defect positioning and identifying module is used for acquiring real-time online ultrasonic detection C-scan data of the carbon fiber composite material to be detected, determining relative position information of the defect, and performing classification identification and quantitative analysis on the defect image by using the trained convolutional neural network.
9. A terminal device comprising a processor and a memory, the processor being arranged to implement instructions; the memory is used for storing a plurality of instructions, wherein the instructions are suitable for being loaded by the processor and executing the carbon fiber composite material defect damage classification identification and quantitative analysis method of any one of claims 1 to 7.
10. A computer-readable storage medium having stored thereon a plurality of instructions, wherein the instructions are adapted to be loaded by a processor of a terminal device and to perform the method for classification, identification and quantitative analysis of defect damage in carbon fiber composite material according to any one of claims 1 to 7.
CN202110271298.XA 2021-03-13 2021-03-13 Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material Pending CN113012123A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110271298.XA CN113012123A (en) 2021-03-13 2021-03-13 Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110271298.XA CN113012123A (en) 2021-03-13 2021-03-13 Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material

Publications (1)

Publication Number Publication Date
CN113012123A true CN113012123A (en) 2021-06-22

Family

ID=76406344

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110271298.XA Pending CN113012123A (en) 2021-03-13 2021-03-13 Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material

Country Status (1)

Country Link
CN (1) CN113012123A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114240947A (en) * 2022-02-28 2022-03-25 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Construction method and device of sweep image database and computer equipment
CN114324582A (en) * 2021-12-10 2022-04-12 哈尔滨工业大学 HHT (Hilbert-Huang transform) and neural network-based composite plate layering defect air coupling ultrasonic detection method
CN115494092A (en) * 2022-10-08 2022-12-20 山东大学 Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network
CN116503615A (en) * 2023-06-25 2023-07-28 南京矩视科技有限公司 Convolutional neural network-based carbon fiber channel identification and numbering method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180075594A1 (en) * 2016-09-14 2018-03-15 Kla-Tencor Corporation Convolutional Neural Network-based Mode Selection and Defect Classification for Image Fusion
CN110363746A (en) * 2019-06-13 2019-10-22 西安交通大学 A kind of Ultrasonic nondestructive test signal classification method based on convolutional neural networks
CN112014476A (en) * 2020-09-04 2020-12-01 山东大学 Method and system for classifying and quantitatively evaluating manufacturing defects of carbon fiber composite material

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180075594A1 (en) * 2016-09-14 2018-03-15 Kla-Tencor Corporation Convolutional Neural Network-based Mode Selection and Defect Classification for Image Fusion
CN110363746A (en) * 2019-06-13 2019-10-22 西安交通大学 A kind of Ultrasonic nondestructive test signal classification method based on convolutional neural networks
CN112014476A (en) * 2020-09-04 2020-12-01 山东大学 Method and system for classifying and quantitatively evaluating manufacturing defects of carbon fiber composite material

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
GENG,X: "Research on FBG-Based CFRP Structural Damage Identification Using BP Neural Network", 《PHOTONIC SENSORS》 *
詹湘琳等: "基于一维CNN的碳纤维复合材料缺陷类型判别", 《激光与光电子学进展》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114324582A (en) * 2021-12-10 2022-04-12 哈尔滨工业大学 HHT (Hilbert-Huang transform) and neural network-based composite plate layering defect air coupling ultrasonic detection method
CN114240947A (en) * 2022-02-28 2022-03-25 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Construction method and device of sweep image database and computer equipment
CN114240947B (en) * 2022-02-28 2022-06-14 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） Construction method and device of sweep image database and computer equipment
CN115494092A (en) * 2022-10-08 2022-12-20 山东大学 Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network
CN116503615A (en) * 2023-06-25 2023-07-28 南京矩视科技有限公司 Convolutional neural network-based carbon fiber channel identification and numbering method
CN116503615B (en) * 2023-06-25 2023-08-25 南京矩视科技有限公司 Convolutional neural network-based carbon fiber channel identification and numbering method

Similar Documents

Publication Publication Date Title
CN113012123A (en) Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material
EP3382386B1 (en) Defect detection using ultrasound scan data
Medak et al. Automated defect detection from ultrasonic images using deep learning
CN112083017A (en) Intelligent detection device, method and medium for internal defects of welding seam
CN110136101B (en) Tire X-ray defect detection method based on twinning distance comparison
Posilović et al. Flaw detection from ultrasonic images using YOLO and SSD
JP2017049974A (en) Discriminator generator, quality determine method, and program
CN111325738B (en) Intelligent detection method and system for transverse hole peripheral cracks
CN111754463B (en) Method for detecting CA mortar layer defects of ballastless track based on convolutional neural network
CN113269718B (en) Concrete prefabricated part crack detection method based on deep learning
CN112816556B (en) Defect detection method, device, equipment and storage medium
CN109584206B (en) Method for synthesizing training sample of neural network in part surface flaw detection
US10775349B2 (en) Defect detection using ultrasound scan data
CN115760837A (en) Crystal quality evaluation method and system based on deep neural network
CN115661572A (en) Casting defect recognition model training method, defect recognition method, device and system
CN113554645A (en) Industrial anomaly detection method and device based on WGAN
Aldrin et al. The need and approach for characterization-US air force perspectives on materials state awareness
CN112816557B (en) Defect detection method, device, equipment and storage medium
CN108765391A (en) A kind of plate glass foreign matter image analysis methods based on deep learning
CN111429441A (en) Crater identification and positioning method based on YO L OV3 algorithm
Filipović et al. Automated ultrasonic testing of materials based on C-scan flaw classification
Rahman et al. Viola-Jones Algorithm for Automatic Detection of Hyperbolic Regions in GPR Profiles of Bridge Decks
Sutcliffe et al. Automatic defect recognition of single-v welds using full matrix capture data, computer vision and multi-layer perceptron artificial neural networks
CN111257422B (en) Wheel axle defect identification model construction method and defect identification method based on machine vision
CN110910373B (en) Identification method of orthotropic steel bridge deck fatigue crack detection image

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20210622

RJ01 Rejection of invention patent application after publication