CN116754484B - Nondestructive testing method for nonmetallic liner fiber winding container - Google Patents

Nondestructive testing method for nonmetallic liner fiber winding container Download PDF

Info

Publication number
CN116754484B
CN116754484B CN202310726873.XA CN202310726873A CN116754484B CN 116754484 B CN116754484 B CN 116754484B CN 202310726873 A CN202310726873 A CN 202310726873A CN 116754484 B CN116754484 B CN 116754484B
Authority
CN
China
Prior art keywords
flaw
defect
apparent
fiber layer
internal
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.)
Active
Application number
CN202310726873.XA
Other languages
Chinese (zh)
Other versions
CN116754484A (en
Inventor
周华
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.)
Special Equipment Safety Supervision Inspection Institute of Jiangsu Province
Original Assignee
Special Equipment Safety Supervision Inspection Institute of Jiangsu Province
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 Special Equipment Safety Supervision Inspection Institute of Jiangsu Province filed Critical Special Equipment Safety Supervision Inspection Institute of Jiangsu Province
Priority to CN202310726873.XA priority Critical patent/CN116754484B/en
Publication of CN116754484A publication Critical patent/CN116754484A/en
Application granted granted Critical
Publication of CN116754484B publication Critical patent/CN116754484B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

The invention discloses a nondestructive testing method for a fiber winding container of a nonmetallic liner, which belongs to the field of nondestructive testing and comprises the following steps: carrying out hierarchical light wave irradiation with different colors and brightness on a detection surface of the pressure container to obtain an internal and apparent hierarchical light wave image set; inputting the image set into a container flaw identification model to obtain an internal and apparent flaw identification set; performing feature aggregation on the two identification sets according to the space position information to obtain an internal and apparent flaw data set; screening the apparent flaw data set to obtain anomaly and fiber layer flaw defect information; performing flaw authentication on the internal flaw data set to obtain flaw defect information of the liner layer; the defect and defect information of the fiber layer and the liner layer form a final defect detection result. The technical problem that flaw defect detection efficiency and detection accuracy are low among the prior art has been solved to this application, reaches the detection efficiency and the technical effect of detection accuracy who improves nonmetal inner bag fiber winding pressure vessel.

Description

Nondestructive testing method for nonmetallic liner fiber winding container
Technical Field
The invention relates to the field of nondestructive testing, in particular to a nondestructive testing method for a nonmetallic liner fiber winding container.
Background
The nonmetal liner fiber winding pressure vessel is widely applied to the fields of chemical industry, aerospace and the like, has a complex structure and a thinner wall thickness, and can directly threaten the use safety once the defect occurs. The existing nondestructive testing method mainly comprises radiation detection, liquid penetration detection, acoustic emission detection, magnetic powder detection and the like, and is difficult to realize efficient and accurate nondestructive evaluation.
Disclosure of Invention
The application provides a nondestructive testing method for a nonmetal liner fiber winding container, and aims to solve the technical problems that in the prior art, flaw defect detection efficiency and detection accuracy of the nonmetal liner fiber winding pressure container are low, and the quality of the nonmetal liner fiber winding container is poor.
In view of the above, the present application provides a nondestructive testing method for a nonmetallic liner fiber wound container.
In a first aspect of the present disclosure, there is provided a non-destructive inspection method of a nonmetallic liner filament wound container, the method comprising: dividing a detection surface of a target pressure container to obtain an apparent detection surface and an internal detection surface; carrying out level light wave irradiation on the apparent detection surface, and carrying out multi-view image acquisition by adopting an image acquisition device to obtain an apparent level light wave image set, wherein the apparent level light wave image set comprises K spliced apparent level light wave images; carrying out hierarchical light wave irradiation on the internal detection surface, and carrying out multi-view image acquisition by adopting an image acquisition device to obtain an internal hierarchical light wave image set, wherein the internal hierarchical light wave image set comprises K spliced internal hierarchical light wave images; respectively inputting the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in a container flaw identification model to obtain an apparent flaw identification set and an internal flaw identification set, wherein flaw identifications have spatial position marks; performing flaw feature aggregation on the apparent flaw identification set and the internal flaw identification set based on the spatial position marks to obtain an apparent flaw data set and an internal flaw data set; obtaining a fiber layer flaw feature set, and screening an apparent flaw data set based on the fiber layer flaw feature set to obtain an abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapped flaw defect information and non-identification flaw defect information; performing flaw authentication on the internal flaw data set based on the abnormal flaw data set to obtain flaw defect information of the liner layer; the fiber layer flaw defect information and the liner layer flaw defect information form a flaw defect identification result of the target pressure vessel.
In another aspect of the present disclosure, there is provided a non-destructive inspection system for a nonmetallic liner filament wound container, the system comprising: the detection surface dividing module is used for dividing the detection surface of the target pressure vessel to obtain an apparent detection surface and an internal detection surface; the system comprises an apparent light wave image module, a multi-view image acquisition module and a display module, wherein the apparent light wave image module is used for carrying out hierarchical light wave illumination on an apparent detection surface and adopting an image acquisition device to acquire an apparent hierarchical light wave image set, and the apparent hierarchical light wave image set comprises K spliced apparent hierarchical light wave images; the internal light wave image module is used for carrying out hierarchical light wave irradiation on the internal detection surface, carrying out multi-view image acquisition by adopting the image acquisition device, and obtaining an internal hierarchical light wave image set, wherein the internal hierarchical light wave image set comprises K spliced internal hierarchical light wave images; the flaw identification acquisition module is used for respectively inputting the apparent level light wave image set and the internal level light wave image set into the first identification module and the second identification module in the container flaw identification model to obtain an apparent flaw identification set and an internal flaw identification set, wherein the flaw identification has a space position mark; the flaw feature aggregation module is used for conducting flaw feature aggregation on the apparent flaw identification set and the internal flaw identification set based on the spatial position marks to obtain an apparent flaw data set and an internal flaw data set; the defect data screening module is used for obtaining a fiber layer defect characteristic set, screening an apparent defect data set based on the fiber layer defect characteristic set, and obtaining an abnormal defect data set and fiber layer defect information, wherein the abnormal defect data set comprises overlapped defect information and non-identification defect information; the internal flaw authentication module is used for carrying out flaw authentication on the internal flaw data set based on the abnormal flaw data set to obtain flaw defect information of the liner layer; the defect identification result module is used for forming a defect identification result of the target pressure container by the defect information of the fiber layer and the defect information of the liner layer.
One or more technical solutions provided in the present application have at least the following technical effects or advantages:
firstly, respectively carrying out hierarchical light wave irradiation with different colors and brightness on an internal detection surface and an apparent detection surface of a target pressure container to obtain an internal hierarchical light wave image set and an apparent hierarchical light wave image set; then, inputting the two image sets into a first recognition module and a second recognition module of a container flaw recognition model to obtain an internal flaw identification set and an apparent flaw identification set; performing feature aggregation on the two identification sets according to the space position information to obtain an internal flaw data set and an apparent flaw data set; further, screening the apparent flaw data set based on the fiber layer flaw feature set to obtain an abnormal flaw data set and fiber layer flaw defect information; and finally, performing flaw authentication on the internal flaw data set through the abnormal flaw data set to obtain flaw defect information of the liner layer. The technical scheme that the fiber layer flaw defect information and the liner layer flaw defect information form the final flaw detection result of the target pressure container realizes the efficient and accurate nondestructive evaluation of the internal structure of the target pressure container through bidirectional optical irradiation and multi-channel image recognition, solves the technical problems of low flaw defect detection efficiency and detection accuracy of the non-metal liner fiber winding pressure container in the prior art, and poor quality of the non-metal liner fiber winding container, and achieves the technical effects of improving the detection efficiency and detection accuracy of the non-metal liner fiber winding pressure container and improving the container quality.
The foregoing description is only an overview of the technical solutions of the present application, and may be implemented according to the content of the specification in order to make the technical means of the present application more clearly understood, and in order to make the above-mentioned and other objects, features and advantages of the present application more clearly understood, the following detailed description of the present application will be given.
Drawings
FIG. 1 is a schematic diagram of a possible flow chart of a nondestructive testing method for a fiber winding container with a nonmetallic liner according to an embodiment of the present application;
FIG. 2 is a schematic flow chart of a possible process for obtaining an apparent flaw identification set and an internal flaw identification set in a nondestructive testing method of a nonmetallic liner fiber winding container according to an embodiment of the present application;
fig. 3 is a schematic flow chart of a possible process for obtaining flaw and defect information of a liner layer in a nondestructive testing method of a non-metallic liner fiber winding container according to an embodiment of the present application;
fig. 4 is a schematic diagram of a possible structure of a nondestructive testing system for a fiber winding container with a nonmetallic liner according to an embodiment of the present application.
Reference numerals illustrate: the device comprises a detection surface dividing module 11, an apparent light wave image module 12, an internal light wave image module 13, a flaw identification acquisition module 14, a flaw characteristic aggregation module 15, a flaw data screening module 16, an internal flaw authentication module 17 and a flaw identification result module 18.
Detailed Description
The technical scheme provided by the application has the following overall thought:
the embodiment of the application provides a nondestructive testing method for a nonmetal liner fiber winding container. Through bidirectional optical irradiation and multi-channel image recognition, efficient and accurate nondestructive evaluation of the internal structure of the target pressure container is realized. Firstly, carrying out hierarchical light wave irradiation with different colors and brightness on an inner detection surface and an apparent detection surface of a target pressure container to obtain an inner hierarchical light wave image set and an apparent hierarchical light wave image set so as to comprehensively obtain optical information of the inner part and the outer part of the target pressure container; then, inputting the two image sets into a first recognition module and a second recognition module of a container flaw recognition model to obtain an internal flaw identification set and an apparent flaw identification set, so as to realize preliminary recognition and positioning of the internal and apparent images; furthermore, feature aggregation is carried out on the two identification sets according to the space position information, so that an internal flaw data set and an apparent flaw data set are obtained, and feature extraction and fusion of a primary identification result are realized; then, screening the apparent flaw data set based on the fiber layer flaw feature set to obtain an abnormal flaw data set and fiber layer flaw defect information, and realizing the refined differentiation and correction of the apparent identification result; and finally, performing flaw authentication on the internal flaw data set through the abnormal flaw data set to obtain flaw defect information of the liner layer, and verifying and correcting an internal identification result. The fiber layer flaw defect information and the liner layer flaw defect information form a final flaw detection result of the target pressure vessel.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
Example 1
As shown in fig. 1, an embodiment of the present application provides a nondestructive testing method for a fiber winding container of a nonmetallic liner, which includes:
step S100: dividing a detection surface of a target pressure container to obtain an apparent detection surface and an internal detection surface;
specifically, the target pressure vessel is a nonmetallic liner fiber wound vessel. And firstly, carrying out three-dimensional scanning on the target pressure vessel by adopting a 3D scanner to obtain three-dimensional model data of the target pressure vessel. And then identifying the positions and the ranges of the fiber winding layer and the liner layer in the model, and dividing the positions and the ranges into an apparent detection surface and an inner detection surface respectively, wherein the apparent detection surface corresponds to the fiber winding layer, and the inner detection surface corresponds to the liner layer.
Meanwhile, in the three-dimensional model of the target pressure container, a three-dimensional coordinate system is established for the three-dimensional model of the target pressure container by taking the central point of the target pressure container as an origin, and the apparent detection surface and the internal detection surface are corresponding in the three-dimensional coordinate system, so that the spatial position identifiers of the apparent detection surface and the internal detection surface are obtained, and the follow-up spatial positioning of the identified flaws is facilitated.
The apparent detection surface and the internal detection surface are obtained by dividing the detection surface of the target container, so that a foundation is provided for realizing the comprehensive nondestructive detection of the container.
Step S200: carrying out level light wave irradiation on the apparent detection surface, and carrying out multi-view image acquisition by adopting an image acquisition device to obtain an apparent level light wave image set, wherein the apparent level light wave image set comprises K spliced apparent level light wave images;
specifically, the apparent test corresponds to the filament wound layer of the target pressure vessel. In order to detect the apparent detection surface, optical detection is implemented by adopting a hierarchical light wave irradiation mode, and an image acquisition device is used for acquiring image data from multiple view angles to obtain an apparent hierarchical light wave image set.
Firstly, in the hierarchical light wave irradiation process, light sources with different wavelengths and different light intensities, such as red light, blue light and white light, are selected to sequentially irradiate the apparent detection surface with different light intensities, and the different light wave irradiation enables flaws on the apparent detection surface to present different image characteristics under different light; then using an image acquisition device such as a digital camera, an industrial camera and the like to shoot the apparent detection surface from different view angles to obtain a plurality of two-dimensional images; and then, splicing the two-dimensional images into K Zhang Biaoguan-level light wave images through an image splicing technology to form an apparent-level light wave image set, wherein the image set records image information of the apparent detection surface under different light wave illumination and visual angles, and a data basis is provided for detecting flaws of the apparent detection surface.
The apparent hierarchical light wave image set is obtained by adopting the technical means of hierarchical light wave irradiation and multi-view image acquisition to carry out optical detection on the apparent detection surface, and the image information of the apparent detection surface under different conditions is recorded, so that important detection data is provided for realizing nondestructive detection.
Step S300: performing hierarchical light wave irradiation on the internal detection surface, and performing multi-view image acquisition by adopting the image acquisition device to obtain an internal hierarchical light wave image set, wherein the internal hierarchical light wave image set comprises K spliced internal hierarchical light wave images;
specifically, the inner detection surface corresponds to the inner liner layer of the target pressure vessel. Similar to the apparent detection surface, for detecting the internal detection surface, optical detection is performed by means of hierarchical light wave irradiation, and image data is acquired from multiple perspectives using an image acquisition device, obtaining an internal hierarchical light wave image set. Firstly, selecting a small-sized laser or LED light source, mounting the small-sized laser or LED light source on a slide rail, placing the small-sized laser or LED light source into the inner container layer through the container opening, and irradiating the inner wall surface with hierarchical light waves by using the light source; then, an image acquisition device such as an endoscope and a pipeline camera is arranged on the sliding rail, and enters the inner container layer along with the light source to acquire multi-view images. And then, acquiring two-dimensional images of a plurality of internal detection surfaces after the internal monitoring surface layer level light wave irradiation and the multi-view image acquisition, splicing the two-dimensional images into K internal level light wave images through an image splicing technology to form an internal level light wave image set, recording image information of the internal detection surfaces under different light wave irradiation and view angles, and providing a data basis for detecting flaws of the internal detection surfaces.
The internal detection surface is detected to obtain an internal level light wave image set, so that information acquisition of the internal detection surface under different illumination conditions is realized, important detection data is provided for realizing nondestructive detection, and meanwhile, light sources with different wavelengths and light intensities are selected for irradiation, so that the detection precision is improved.
Step S400: respectively inputting the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in a container flaw identification model to obtain an apparent flaw identification set and an internal flaw identification set, wherein flaw identifications have space position marks;
specifically, the obtained apparent level light wave image set and the internal level light wave image set contain image information of the apparent detection surface and the internal detection surface of the target pressure vessel under different illumination conditions. In order to identify flaw information in the image set, a first identification module and a second identification module in a container flaw identification model are adopted for image identification. The first recognition module corresponds to the apparent detection surface and inputs an apparent level light wave image set; the second recognition module corresponds to the internal detection surface and inputs an internal-level light wave image set. The two recognition modules are obtained through training of a deep learning technology and are used for recognizing the types and positions of flaws in the corresponding detection surfaces.
After the apparent level light wave image set and the internal level light wave image set are input into two recognition modules, the modules respectively recognize and detect the input image sets. The method comprises the steps that a first identification module extracts images in an apparent level light wave image set one by one, detects flaws in a detection surface corresponding to the apparent level light wave image, generates apparent flaw identification, integrates a three-dimensional model, endows flaw space position marks, and detects all the images in the apparent level light wave image set to obtain an apparent flaw identification set, wherein the images in the apparent level light wave image set correspond to flaw identification mapping in the apparent flaw identification set; and the second recognition module extracts images in the internal-level light wave image set one by one, detects flaws in a detection surface corresponding to the internal-level light wave image, generates an internal flaw mark, integrates a three-dimensional model to endow flaw space position marks, detects all the images in the internal-level light wave image set, and obtains an internal flaw mark set, wherein the images in the internal-level light wave image set correspond to flaw mark mapping in the internal flaw mark set.
The apparent level light wave image set and the internal level light wave image set are detected by utilizing the two recognition modules of the container flaw recognition model respectively, the apparent flaw identification set and the internal flaw identification set are generated, and flaw identifications are provided with spatial position marks, so that mapping with the input light wave image set is realized, and finally, the flaw identifications are mapped to a three-dimensional coordinate system of a container, and important spatial information support is provided for judging the defect condition of the container and improving the detection precision.
Step S500: performing flaw feature aggregation on the apparent flaw identification set and the internal flaw identification set based on the spatial position marks to obtain an apparent flaw data set and an internal flaw data set;
specifically, the apparent flaw identification set and the internal flaw identification set contain all flaw identifications and spatial position information detected in the apparent detection surface and the internal detection surface. To further aggregate flaw features and determine container quality, both sets of signatures are processed.
Firstly, extracting the space position information of each flaw mark in the apparent flaw mark set and the internal flaw mark set, and representing the space coordinates of the flaw mark on a container; then, calculating the distance or the overlapping area between the spatial positions of any two flaw marks, and judging that the two flaws belong to the same flaw feature if the spatial positions of the two flaw marks are close (the distance is smaller than a preset distance threshold) or overlap (the overlapping area is larger than an overlapping area threshold); then, according to the judged association relation of the flaw features, flaw marks belonging to the same feature are aggregated together to form an apparent flaw data set or an internal flaw data set, and each data set represents a flaw feature; repeating the extraction of one flaw identification from each of the apparent flaw identification set and the internal flaw identification set, and judging the spatial correlation between any two flaw identifications in the apparent flaw identification set and the internal flaw identification set so as to realize the aggregation of all flaw features; finally, apparent flaw data sets and internal flaw data sets are obtained, wherein each data set contains all flaw identifications belonging to the same flaw feature, and detailed information of the feature is recorded, including flaw types, quantity, spatial range and the like.
And judging the spatial correlation between flaw marks according to the spatial position marks in the apparent flaw mark set and the internal flaw mark set, and realizing flaw feature aggregation. And aggregating flaw identifications belonging to the same flaw feature to generate an apparent flaw data set and an internal flaw data set. Each data set represents a flaw signature that provides aggregated flaw information for determining container quality.
Step S600: obtaining a fiber layer flaw feature set, and screening the apparent flaw data set based on the fiber layer flaw feature set to obtain an abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapped flaw defect information and non-identification flaw defect information;
specifically, in order to determine whether the flaw features in the apparent flaw data set belong to a fiber layer flaw in the material property of the container itself, a fiber layer flaw feature set is obtained for comparison. The fiber layer flaw feature set is obtained through historical experience extraction and summarization, and comprises common fiber layer flaw features of pressure vessels made of different materials, including flaw type information, characteristic parameters and the like.
The apparent flaw dataset is screened based on the fiber layer flaw feature set. And if the characteristic of a certain flaw in the apparent flaw data set is matched with the characteristic in the fiber layer flaw characteristic set, judging that the flaw belongs to the fiber layer flaw of the container material. The matched flaw features are extracted to form flaw defect information of the fiber layer, and the flaw defect information is the flaw state of the fiber layer of the container. If a flaw feature in the apparent flaw data set does not find a matching item in the fiber layer flaw feature set, judging that the flaw belongs to an abnormal flaw, and overlapping the flaw or an unrecognizable flaw. These anomaly blemishes are extracted to form an anomaly blemish dataset.
By obtaining a fiber layer flaw feature set, the apparent flaw data set is screened based on the set. And extracting the matched flaw features to form flaw defect information of the fiber layer. The unmatched abnormal flaws are extracted to form an abnormal flaw data set, so that the quality of the container can be judged, and the detection accuracy of the nonmetal liner fiber winding pressure container can be improved.
Step S700: performing flaw authentication on the internal flaw data set based on the abnormal flaw data set to obtain flaw defect information of the liner layer;
specifically, the abnormal flaw data set contains unrecognized flaw or overlapping flaw information detected in the apparent detection surface. The internal flaw dataset contains flaw signature information detected in the internal detection face. To determine whether the defect information in the abnormal defect data set also corresponds to the internal detection surface, defect authentication is performed on the internal defect data set. And extracting each piece of flaw information in the abnormal flaw data set, and searching a matching item in the internal flaw data set. If the matched flaw features are found, judging that the flaws simultaneously correspond to the apparent detection surface and the internal detection surface, and belong to flaws of the liner layer. And extracting all the matched flaws found in the internal flaw data set to form flaw defect information of the liner layer, wherein the flaw defect information is flaws and related parameters of the liner layer of the container.
And performing flaw authentication on the internal flaw data set based on the abnormal flaw data set, judging flaws belonging to the liner layer if the internal flaw data set contains matched flaw features, extracting flaw defect information forming the liner layer, accurately judging flaws of the liner layer, and providing a reliable basis for judging the quality of the detection container.
Step S800: and the fiber layer flaw defect information and the liner layer flaw defect information form a flaw defect identification result of the target pressure container.
Specifically, the defect information of the fiber layer comprises various defect characteristics and parameters of the fiber layer of the container, such as the number, type, position, size and the like of the defects. The defect information of the liner layer comprises various defect characteristics and parameters of the liner layer of the container. Both of these information include information about imperfections present on the surface and inside the container. And combining the defect information of the fiber layer and the defect information of the liner layer to be used as a defect and defect identification result so as to comprehensively understand the defect condition of the container, the defect source and the type and provide detailed information basis for judging the quality of the container.
The defect information of the fiber layer and the defect information of the liner layer form the defect and defect identification result of the target pressure container, the information of the fiber layer defects and the liner layer defects is extracted, the defect condition of the container is comprehensively reflected, and the technical effects of improving the detection efficiency and the detection accuracy of the nonmetallic liner fiber winding pressure container and improving the quality of the container are achieved.
Further, as shown in fig. 2, the embodiment of the present application further includes:
step S410: interactively obtaining a target model parameter of the target pressure container;
step S420: extracting and obtaining historical flaw detection information based on the target model parameters, wherein the historical flaw detection information comprises a sample fiber layer flaw image set and a sample liner layer flaw image set;
step S430: dividing the flaw images and the background images of the flaw image set of the sample fiber layer and the flaw image set of the sample liner layer based on semantic division to obtain a flaw image set of a training sample;
step S440: performing synchronous construction of the first recognition module and the second recognition module based on the training sample flaw image set to generate the container flaw recognition model;
step S450: and respectively inputting the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in the container flaw identification model to obtain the apparent flaw identification set and the internal flaw identification set.
Specifically, the target model parameters of the target pressure vessel are input and obtained through a human-computer interaction interface. The target model parameter contains information such as materials, structures, volumes and the like of the container, and provides an important basis for extracting historical flaw detection information. And recording various flaw information of the container with the same or similar model parameters in the using process, searching information matched with the flaw information in a historical flaw detection information base based on the input target model parameters to form historical flaw detection information, wherein the historical flaw detection information comprises image information obtained for an apparent detection surface and an internal detection surface to respectively form a sample fiber layer flaw image set and a sample liner layer flaw image set.
The sample fiber layer flaw image set and the sample liner layer flaw image set contain various flaw images of the target container, but also contain a large amount of background information. In order to obtain a relatively accurate training sample, two sample sets are required to be subjected to semantic segmentation, flaw targets and backgrounds are divided from sample images, the flaw targets are extracted to form a training sample flaw image set which comprises a fiber layer flaw image and a liner layer flaw image, and ideal input is provided for training a container flaw identification model.
A convolutional neural network is selected as a construction algorithm of a first recognition module and a second recognition module; and dividing the training sample flaw image set for training of the first identification module and the second identification module. Inputting the divided images into corresponding identification modules, and synchronously starting the training process by adopting the selected convolutional neural network. The first recognition module is used for recognizing apparent flaws, and training output is the category of the apparent flaws; the second recognition module is used for recognizing internal flaws, and training output is the internal flaw category.
The apparent hierarchical lightwave image set is from apparent detection facing detection of the target container, and the internal hierarchical lightwave image set is from internal detection facing detection of the target container; inputting the apparent level light wave image set into a first identification module, and identifying apparent flaws to obtain an apparent flaw identification set; and inputting the internal hierarchical light wave image set into a second identification module, so that internal flaws can be identified, and an internal flaw identification set is obtained.
The model parameters of the target container are obtained, history flaw detection information matched with the model parameters is extracted, the model parameters comprise a sample fiber layer flaw image set and a sample liner layer flaw image set, semantic segmentation is carried out on the two sample sets, flaw images are extracted to form a training sample flaw image set, a first identification module and a second identification module are synchronously constructed based on the sample sets, a container flaw identification model is finally generated, an apparent level light wave image set and an internal level light wave image set are input into the identification model, an apparent flaw identification set and an internal flaw identification set are respectively obtained, automatic identification of images of an apparent detection surface and an internal detection surface of the container is achieved, and an important basis is provided for subsequent judgment of the flaw.
Further, the embodiment of the application further includes:
step S510: performing flaw space positioning on the apparent flaw identification set based on the space position mark to obtain the apparent flaw data set, wherein the apparent flaw data set comprises H groups of overlapped apparent flaw image sets, and H is a positive integer;
step S520: performing flaw space positioning on the internal flaw identification set based on the space position mark to obtain the internal flaw data set, wherein the internal flaw data set comprises N groups of overlapped internal flaw image sets, and N is a positive integer;
Specifically, each flaw identification in the set of apparent flaw identifications is associated with three-dimensional spatial coordinates on the apparent detection surface based on the spatial location markers. All the marks are spatially positioned to form an apparent flaw data set. Because the apparent detection surface adopts a multi-level light wave imaging technology, a flaw can be identified for a plurality of times to generate overlapped flaw images, the apparent flaw data set is divided into H groups, each group comprises flaw image sets from one layer of light wave detection, and H is the number of layers of multi-level light wave imaging. The apparent flaw data set records flaw information and images existing on the apparent detection surface, and provides important input for extracting apparent layer flaw information.
Based on the spatial location markers, each flaw marker in the set of internal flaw markers is associated with three-dimensional spatial coordinates on the internal detection surface. All the marks are spatially positioned to form an internal flaw data set. Because the internal detection surface also adopts a multi-level light wave imaging technology, a flaw can be identified for multiple times to generate overlapped flaw images, the internal flaw data set is divided into N groups, each group comprises flaw image sets from one-layer light wave detection, and N is the number of layers of multi-level light wave imaging. The internal flaw data set records flaw information and images existing on the internal detection surface, and provides important input for extracting flaw information of the liner layer.
The flaw data is spatially positioned in a three-dimensional coordinate system of the target pressure vessel model, multi-layer light wave imaging is divided, the apparent flaw identification set and the internal flaw identification set are converted into an apparent flaw data set and an internal flaw data set, and judgment of flaws is provided for a more detailed basis.
Further, the embodiment of the application further includes:
step S610: performing flaw feature extraction based on the sample fiber layer flaw image set to obtain the fiber layer flaw feature set;
step S620: extracting and obtaining a first overlapped apparent flaw image set based on the H groups of overlapped apparent flaw image sets;
step S630: judging whether the first overlapped apparent flaw image set meets the fiber layer flaw feature set or not;
step S640: if the first overlapped apparent flaw image set meets the fiber layer flaw feature set, extracting feature images of the first overlapped apparent image set to obtain a first fiber layer flaw defect;
step S650: adding the first fiber layer flaw defect into the fiber layer flaw defect information;
step S660: if part of apparent flaw images in the first overlapped apparent flaw image set meet the fiber layer flaw feature set, multi-feature image extraction is carried out on the first overlapped apparent image set, and a first overlapped flaw defect is obtained;
Step S670: adding the first overlapped flaw defect to the overlapped flaw defect information;
step S680: if the first overlapped apparent flaw image set does not meet the fiber layer flaw feature set, extracting feature images of the first overlapped apparent image set to obtain a first non-identification flaw defect;
step S690: adding the first unidentified flaw defect to the unidentified flaw defect information;
step S6100: and by analogy, comparing the fiber layer flaw characteristic set with the H groups of overlapping apparent flaw image sets to obtain the abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapping flaw defect information and non-identification flaw defect information.
Specifically, the sample fiber layer defect image set contains a large number of typical fiber layer defect images, and the images are analyzed by using an image processing technology to extract characteristics, such as shape, color, texture and the like, which can represent fiber layer defects, so as to form a fiber layer defect characteristic set. For example, the shape characteristics such as the perimeter, the area, the axial length ratio, the rectangularity and the like of the flaw area are extracted through the shape characteristics; calculating color characteristics such as average RGB value, dominant RGB value, chromaticity and the like of the flaw area through color characteristic extraction; and calculating texture features such as roughness, contrast, variance and the like of the flaw area through texture feature extraction.
And extracting a group of overlapped apparent flaw image sets from the H groups of overlapped apparent flaw image sets for judgment, wherein the first overlapped apparent flaw image set comprises all flaw images obtained by apparent detection. And comparing the first overlapped apparent flaw image set with the fiber layer flaw feature set, and judging whether flaw features in the image set are matched with the fiber layer flaw feature set. The result of the matching may determine whether the first overlaid apparent defect image set belongs to a fibrous layer defect.
If the comparison result shows that the first overlapped apparent flaw image set belongs to a fiber layer flaw, extracting a characteristic image representing the flaw from the first overlapped apparent flaw image set to form a first fiber layer flaw, wherein the first fiber layer flaw comprises detailed information such as the spatial position, the category and the like of the flaw. And adding the first fiber layer defect into the fiber layer defect information, and recording the first fiber layer defect and other fiber layer defect together to provide a basis for finally judging the quality of the container.
If the comparison result shows that only part of the first overlapped apparent flaw image set belongs to the fiber layer flaw, extracting a plurality of characteristic images from the image set, and respectively representing different flaw categories to form a first overlapped flaw defect. The first overlapped flaw defect is added into the overlapped flaw defect information and recorded together with other overlapped flaw defects, so that a reference is provided for finally judging the quality of the container.
If the comparison result of the first overlapped apparent flaw image set and the fiber layer flaw feature set shows that the image set does not belong to the fiber layer flaw, the feature image representing the flaw needs to be extracted from the image set to form a first non-identification flaw defect. The first overlapped flaw defect is added into the overlapped flaw defect information and recorded together with other overlapped flaw defects, so that a reference is provided for finally judging the quality of the container.
And by analogy, comparing the fiber layer flaw feature set with the H-group overlapped apparent flaw image set of the apparent flaw data set group by group to obtain an abnormal flaw data set and fiber layer flaw defect information. Recording defect information which does not belong to defects of the fiber layer by using the abnormal defect data set; the defect information of the fiber layer defect records detailed information belonging to the fiber layer defect.
By intelligently judging and processing a large number of images obtained by imaging the apparent detection surface light waves, the automatic identification of defects of the fiber layer is realized, the difficulty of manual judgment is greatly reduced, the detection efficiency of a target pressure container is improved, and a foundation is provided for screening apparent defect data sets.
Further, as shown in fig. 3, the embodiment of the present application further includes:
Step S710: performing flaw feature extraction on the flaw image set of the sample liner layer to obtain a flaw feature set of the liner layer;
step S720: performing flaw authentication of the non-identification flaw defect information based on the flaw feature set of the liner layer to obtain a flaw defect set of the first liner layer;
step S730: screening the overlapped flaw defect information according to the flaw feature set of the liner layer and the flaw feature set of the fiber layer to obtain a flaw defect set of the second liner layer and flaw defect compensation of the fiber layer;
step S740: performing single characteristic identification on the N groups of overlapped internal flaw image sets through the flaw characteristic sets of the inner liner layer to obtain a flaw defect set of a third inner liner layer;
step S750: performing spatial position aggregation on the first liner layer flaw defect set, the second liner layer flaw defect set and the third liner layer flaw defect set to obtain liner layer flaw defect information;
step S760: and carrying out data optimization on the defect information of the fiber layer defects based on the defect compensation of the fiber layer defects.
Specifically, the sample liner layer flaw image set includes typical liner layer flaw images, and features representing liner layer flaws, such as shapes, colors, textures and the like, are extracted by using an image processing technology to form a liner layer flaw feature set. And comparing the non-identification flaw defect information with the flaw feature set of the liner layer, and judging whether the flaw feature not belonging to the fiber layer accords with the flaw feature set of the liner layer or not, wherein the information belonging to the flaw of the liner layer forms a first flaw defect set of the liner layer.
The overlapped flaw defect information comprises flaw information of different layers, the flaw information is screened according to the flaw feature set of the liner layer and the flaw feature set of the fiber layer, the overlapped flaw defect information is judged, and the overlapped flaw defect information is divided into flaw information of the liner layer and flaw information of the fiber layer. The information belonging to the flaws of the liner layer constitutes a second liner layer flaw defect set, and the information belonging to the flaws of the fiber layer, which are misjudged as flaws of the liner layer, constitutes fiber layer flaw defect compensation. The internal flaw data set comprises N groups of overlapped internal flaw image sets, each group is from one-layer internal detection, repeated internal flaw data are removed, the N groups of overlapped internal flaw image sets are ensured not to be repeated, and information belonging to flaws of the liner layer is formed into a third liner layer flaw defect set.
The flaw position information in different sets is unified with a container model to be constructed under a three-dimensional coordinate system for the first inner container layer flaw defect set, the second inner container layer flaw defect set and the third inner container layer flaw defect set; and clustering flaw position information by adopting a DBSCAN algorithm and a mean shift clustering algorithm, wherein the information belonging to the same flaw area is gathered into one category, so that flaw defect information of the liner layer is obtained, and detailed spatial position and category information of flaws belonging to the liner layer are recorded.
And finally, optimizing the defect information of the fiber layer defect by utilizing the defect compensation of the fiber layer defect, adding the information misjudged as the defect of the liner layer, improving the judgment accuracy, and taking the optimized defect information of the fiber layer defect and the defect information of the liner layer as the basis for judging the quality of the container.
And extracting and judging the characteristics of the non-identification flaw defect information, the overlapping flaw defect information and the internal flaw image set to obtain flaw defect information of the liner layer. Meanwhile, the defect compensation of the fiber layer is adopted to conduct data optimization, so that the judgment accuracy is improved, and a foundation is provided for improving the detection efficiency and the detection accuracy of the nonmetal liner fiber winding pressure container.
Further, the embodiment of the application further includes:
step S910: the fiber layer defect information comprises M fiber layer defect defects, wherein each fiber layer defect has the spatial position mark;
step S920: performing region division on the apparent detection surface based on the spatial position mark to obtain an apparent detection region division result;
step S930: carrying out hierarchical light intensity irradiation on the apparent detection surface, and adopting the image acquisition device to acquire multi-view images so as to obtain a fiber layer light intensity image set;
Step S940: mapping the apparent detection region division result to the fiber layer light intensity image set for division to obtain an apparent detection region image set;
step S950: performing flaw size extraction based on the apparent detection area image set to obtain M groups of fiber layer flaw size parameters;
step S960: and carrying out serialization treatment on M groups of fiber layer flaw size parameters to obtain M pieces of fiber layer flaw size information.
Specifically, the fiber layer defect information includes M fiber layer defect defects belonging to a fiber layer, wherein the fiber layer defect defects have corresponding spatial position information for identifying specific spatial positions of each defect. And carrying out region division on the surface detection surface according to the spatial position mark in the defect information of the fiber layer, wherein each region corresponds to the spatial range of a defect.
And controlling the light intensity of each region in different levels according to the region division result, and acquiring multi-view images by adopting an image acquisition device to obtain a fiber layer light intensity image set containing detailed information in the region. The hierarchical control of the light intensity can highlight the tiny flaw information in the area, and the detection precision is improved. And mapping the apparent detection region division result to a fiber layer light intensity image set, and dividing images belonging to the same region together to obtain an apparent detection region image set. Each region image set corresponds to an apparent detection region and contains detailed structural information of the region. And (3) performing image processing on each apparent detection area image set, extracting characteristic parameters representing flaw size, such as area, perimeter, main shaft length ratio and the like, and forming M groups of fiber layer flaw size parameters. And sequentially corresponding M groups of fiber layer flaw size parameters to flaw defects of each fiber layer to form M pieces of fiber layer flaw size information, wherein the fiber layer flaw size information records the specific size of flaw defects of each fiber layer.
The fiber layer flaw size information is obtained through apparent detection surface area division, light intensity image acquisition and flaw size extraction based on fiber layer flaw defect information, three-dimensional imaging of a microscopic structure of the apparent detection surface is realized, and a quantification basis is provided for judging the quality of the fiber layer, so that high-precision nondestructive detection is realized.
Further, the embodiment of the application further includes:
step S971: and interactively obtaining a flaw size constraint set of the fiber layer flaw feature set, wherein the flaw size constraint set has a mapping relation with the fiber layer flaw feature set.
Step S972: performing defect type identification of the M fiber layer defect defects based on the fiber layer defect feature set to obtain M fiber layer defect type identifiers;
step S973: mapping and extracting the flaw size constraint set based on the M fiber layer flaw type identifiers to obtain M flaw size constraints;
step S974: judging whether the M fiber layer flaw size information completely meets the M flaw size constraints;
step S975: if the M flaw size information of the fiber layers completely meets the M flaw size constraints, generating a fiber layer flaw unobstructed instruction;
Step S976: and if any one of the M fiber layer flaw size information does not meet the M flaw size constraints, generating a pressure vessel scrapping instruction.
Specifically, firstly, through collecting relevant historical detection data, characteristic information of various fiber layer flaws and a determined allowable size range are counted, a fiber layer flaw characteristic set and a flaw size constraint set are constructed, and a mapping relation between the two is established. And secondly, matching the characteristic information of each defect with a fiber layer flaw characteristic set, and finding out the most similar flaw type to obtain a corresponding fiber layer flaw type mark. And thirdly, calculating the matching degree or similarity between each mark and each size constraint in the flaw size constraint set, and extracting the size constraint with the highest matching degree to obtain M flaw size constraints. And then comparing the obtained M pieces of fiber layer flaw size information with M pieces of flaw size constraints, and judging whether the actual size of each fiber layer flaw defect is in an allowable range or not so as to judge whether all the constraints are met or not. If the actual sizes of the M fiber layer flaw defects all meet the corresponding size constraint, the quality of the fiber layer is considered to meet the requirement, and an unimpeded instruction of the fiber layer flaw is generated, so that the quality of the fiber layer reaches the standard. If any one of the actual sizes of the M fiber layer flaw defects exceeds the corresponding size constraint, the quality of the fiber layer is considered to be unsatisfactory, and a pressure container rejection instruction is generated. Indicating that the quality of the fiber layer does not reach the standard, and the scrapping treatment is needed.
The defect type identification and the size judgment of the fiber layer defect feature set and the actual acquired information finally generate an unimpeded instruction or a scrapped instruction, so that the quantitative judgment of a detection result is realized, the high-precision nondestructive detection is realized, the detection efficiency and the detection accuracy of the nonmetal liner fiber winding pressure container are improved, and the technical effect of improving the container quality is achieved.
In summary, the nondestructive testing method for the nonmetallic liner fiber winding container provided by the embodiment of the application has the following technical effects:
the method comprises the steps of dividing a detection surface of a target pressure container to obtain an apparent detection surface and an internal detection surface, realizing accurate division of an internal and external structure of the target pressure container, and providing a basis for subsequent optical information acquisition; carrying out level light wave irradiation on the apparent detection surface, and carrying out multi-view image acquisition by adopting an image acquisition device to obtain an apparent level light wave image set, wherein the apparent level light wave image set comprises K spliced apparent level light wave images; carrying out hierarchical light wave irradiation on the internal detection surface, and carrying out multi-view image acquisition by adopting an image acquisition device to obtain an internal hierarchical light wave image set, wherein the internal hierarchical light wave image set comprises K spliced internal hierarchical light wave images, so that comprehensive optical information acquisition on the internal and external structures of the target pressure vessel is realized; respectively inputting the apparent level light wave image set and the internal level light wave image set into a first recognition module and a second recognition module in a container flaw recognition model to obtain an apparent flaw identification set and an internal flaw identification set, wherein flaw identifications have spatial position marks, and primary recognition and positioning of internal and external images are realized; performing flaw feature aggregation on the apparent flaw identification set and the internal flaw identification set based on the spatial position marks to obtain an apparent flaw data set and an internal flaw data set, so that feature extraction and fusion of a primary identification result are realized; obtaining a fiber layer flaw feature set, and screening an apparent flaw data set based on the fiber layer flaw feature set to obtain an abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapped flaw defect information and non-identification flaw defect information, so that the precise distinguishing and correction of the apparent identification result are realized; performing flaw authentication on the internal flaw data set based on the abnormal flaw data set to obtain flaw defect information of the liner layer, so that verification and correction of an internal identification result are realized; the fiber layer flaw defect information and the liner layer flaw defect information form a flaw defect identification result of the target pressure container, so that final evaluation and judgment of the internal structure of the target pressure container are realized, and the technical effects of improving the detection efficiency and detection accuracy of the nonmetallic liner fiber winding pressure container and improving the container quality are achieved.
Obtaining a fiber layer flaw feature set, and screening an apparent flaw data set based on the fiber layer flaw feature set to obtain an abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapped flaw defect information and non-identification flaw defect information, so that the precise distinguishing and correction of the apparent identification result are realized; performing flaw authentication on the internal flaw data set based on the abnormal flaw data set to obtain flaw defect information of the liner layer, so that verification and correction of an internal identification result are realized; the fiber layer flaw defect information and the liner layer flaw defect information form a flaw defect identification result of the target pressure container, so that final evaluation and judgment of the internal structure of the target pressure container are realized, and the technical effects of improving the detection efficiency and detection accuracy of the nonmetallic liner fiber winding pressure container and improving the container quality are achieved.
Example two
Based on the same inventive concept as the nondestructive testing method of a nonmetallic liner fiber winding container in the foregoing embodiment, as shown in fig. 4, an embodiment of the present application provides a nondestructive testing system of a nonmetallic liner fiber winding container, which includes:
The detection surface dividing module 11 is used for dividing the detection surface of the target pressure vessel to obtain an apparent detection surface and an internal detection surface;
the apparent light wave image module 12 is used for carrying out level light wave irradiation on the apparent detection surface and carrying out multi-view image acquisition by adopting an image acquisition device to obtain an apparent level light wave image set, wherein the apparent level light wave image set comprises K spliced apparent level light wave images;
an internal light wave image module 13, configured to perform hierarchical light wave illumination on the internal detection surface, and perform multi-view image acquisition by using the image acquisition device, so as to obtain an internal hierarchical light wave image set, where the internal hierarchical light wave image set includes K spliced internal hierarchical light wave images;
a flaw identification acquisition module 14, configured to input the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in a container flaw identification model respectively, to obtain an apparent flaw identification set and an internal flaw identification set, where the flaw identification has a spatial position mark;
a flaw feature aggregation module 15, configured to aggregate flaw features of the apparent flaw identification set and the internal flaw identification set based on the spatial location markers, to obtain an apparent flaw data set and an internal flaw data set;
A defect data screening module 16, configured to obtain a fiber layer defect feature set, and screen the apparent defect data set based on the fiber layer defect feature set to obtain an abnormal defect data set and fiber layer defect information, where the abnormal defect data set includes overlapping defect information and non-identifying defect information;
the internal flaw authentication module 17 performs flaw authentication of the internal flaw data set based on the abnormal flaw data set to obtain flaw defect information of the liner layer;
and the defect recognition result module 18 is used for forming a defect recognition result of the target pressure container by the defect information of the fiber layer and the defect information of the liner layer.
Further, the flaw identification acquisition module 14 includes the following steps:
interactively obtaining a target model parameter of the target pressure container;
extracting and obtaining historical flaw detection information based on the target model parameters, wherein the historical flaw detection information comprises a sample fiber layer flaw image set and a sample liner layer flaw image set;
dividing the flaw images and the background images of the flaw image set of the sample fiber layer and the flaw image set of the sample liner layer based on semantic division to obtain a flaw image set of a training sample;
Performing synchronous construction of the first recognition module and the second recognition module based on the training sample flaw image set to generate the container flaw recognition model;
and respectively inputting the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in the container flaw identification model to obtain the apparent flaw identification set and the internal flaw identification set.
Further, the flaw feature aggregation module 15 includes the following steps:
performing flaw space positioning on the apparent flaw identification set based on the space position mark to obtain the apparent flaw data set, wherein the apparent flaw data set comprises H groups of overlapped apparent flaw image sets, and H is a positive integer;
performing flaw space positioning on the internal flaw identification set based on the space position mark to obtain the internal flaw data set, wherein the internal flaw data set comprises N groups of overlapped internal flaw image sets, and N is a positive integer;
further, the flaw data screening module 16 includes the following steps:
performing flaw feature extraction based on the sample fiber layer flaw image set to obtain the fiber layer flaw feature set;
Extracting and obtaining a first overlapped apparent flaw image set based on the H groups of overlapped apparent flaw image sets;
judging whether the first overlapped apparent flaw image set meets the fiber layer flaw feature set or not;
if the first overlapped apparent flaw image set meets the fiber layer flaw feature set, extracting feature images of the first overlapped apparent image set to obtain a first fiber layer flaw defect;
adding the first fiber layer flaw defect into the fiber layer flaw defect information;
if part of apparent flaw images in the first overlapped apparent flaw image set meet the fiber layer flaw feature set, multi-feature image extraction is carried out on the first overlapped apparent image set, and a first overlapped flaw defect is obtained;
adding the first overlapped flaw defect to the overlapped flaw defect information;
if the first overlapped apparent flaw image set does not meet the fiber layer flaw feature set, extracting feature images of the first overlapped apparent image set to obtain a first non-identification flaw defect;
adding the first unidentified flaw defect to the unidentified flaw defect information;
and by analogy, comparing the fiber layer flaw characteristic set with the H groups of overlapping apparent flaw image sets to obtain the abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapping flaw defect information and non-identification flaw defect information.
Further, the internal flaw authentication module 17 includes the following steps:
performing flaw feature extraction on the flaw image set of the sample liner layer to obtain a flaw feature set of the liner layer;
performing flaw authentication of the non-identification flaw defect information based on the flaw feature set of the liner layer to obtain a flaw defect set of the first liner layer;
screening the overlapped flaw defect information according to the flaw feature set of the liner layer and the flaw feature set of the fiber layer to obtain a flaw defect set of the second liner layer and flaw defect compensation of the fiber layer;
performing single characteristic identification on the N groups of overlapped internal flaw image sets through the flaw characteristic sets of the inner liner layer to obtain a flaw defect set of a third inner liner layer;
performing spatial position aggregation on the first liner layer flaw defect set, the second liner layer flaw defect set and the third liner layer flaw defect set to obtain liner layer flaw defect information;
and carrying out data optimization on the defect information of the fiber layer defects based on the defect compensation of the fiber layer defects.
Further, the embodiment of the application further comprises a fiber layer flaw size information module, which comprises the following implementation steps:
The fiber layer defect information comprises M fiber layer defect defects, wherein each fiber layer defect has the spatial position mark;
performing region division on the apparent detection surface based on the spatial position mark to obtain an apparent detection region division result;
carrying out hierarchical light intensity irradiation on the apparent detection surface, and adopting the image acquisition device to acquire multi-view images so as to obtain a fiber layer light intensity image set;
mapping the apparent detection region division result to the fiber layer light intensity image set for division to obtain an apparent detection region image set;
performing flaw size extraction based on the apparent detection area image set to obtain M groups of fiber layer flaw size parameters;
and carrying out serialization treatment on M groups of fiber layer flaw size parameters to obtain M pieces of fiber layer flaw size information.
Further, the fiber layer flaw size information module further includes the following steps:
and interactively obtaining a flaw size constraint set of the fiber layer flaw feature set, wherein the flaw size constraint set has a mapping relation with the fiber layer flaw feature set.
Performing defect type identification of the M fiber layer defect defects based on the fiber layer defect feature set to obtain M fiber layer defect type identifiers;
Mapping and extracting the flaw size constraint set based on the M fiber layer flaw type identifiers to obtain M flaw size constraints;
judging whether the M fiber layer flaw size information completely meets the M flaw size constraints;
if the M flaw size information of the fiber layers completely meets the M flaw size constraints, generating a fiber layer flaw unobstructed instruction;
and if any one of the M fiber layer flaw size information does not meet the M flaw size constraints, generating a pressure vessel scrapping instruction.
Any of the steps of the methods described above may be stored as computer instructions or programs in a non-limiting computer memory and may be called by a non-limiting computer processor to identify any of the methods to implement embodiments of the present application, without unnecessary limitations.
Further, the first or second element may not only represent a sequential relationship, but may also represent a particular concept, and/or may be selected individually or in whole among a plurality of elements. It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the present application and the equivalents thereof, the present application is intended to cover such modifications and variations.

Claims (4)

1. A method for non-destructive testing of a non-metallic liner filament wound container, the method comprising:
dividing a detection surface of a target pressure container to obtain an apparent detection surface and an internal detection surface;
carrying out level light wave irradiation on the apparent detection surface, and carrying out multi-view image acquisition by adopting an image acquisition device to obtain an apparent level light wave image set, wherein the apparent level light wave image set comprises K spliced apparent level light wave images;
performing hierarchical light wave irradiation on the internal detection surface, and performing multi-view image acquisition by adopting the image acquisition device to obtain an internal hierarchical light wave image set, wherein the internal hierarchical light wave image set comprises K spliced internal hierarchical light wave images;
inputting the apparent level light wave image set and the internal level light wave image set into a first recognition module and a second recognition module in a container flaw recognition model respectively to obtain an apparent flaw identification set and an internal flaw identification set, wherein flaw identification has a spatial position mark, and the method comprises the following steps:
interactively obtaining a target model parameter of the target pressure container;
extracting and obtaining historical flaw detection information based on the target model parameters, wherein the historical flaw detection information comprises a sample fiber layer flaw image set and a sample liner layer flaw image set;
Dividing the flaw images and the background images of the flaw image set of the sample fiber layer and the flaw image set of the sample liner layer based on semantic division to obtain a flaw image set of a training sample;
performing synchronous construction of the first recognition module and the second recognition module based on the training sample flaw image set to generate the container flaw recognition model;
respectively inputting the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in the container flaw identification model to obtain the apparent flaw identification set and the internal flaw identification set;
performing flaw feature aggregation on the apparent flaw identification set and the internal flaw identification set based on the spatial position marks to obtain an apparent flaw data set and an internal flaw data set, including:
performing flaw space positioning on the apparent flaw identification set based on the space position mark to obtain the apparent flaw data set, wherein the apparent flaw data set comprises H groups of overlapped apparent flaw image sets, and H is a positive integer;
performing flaw space positioning on the internal flaw identification set based on the space position mark to obtain the internal flaw data set, wherein the internal flaw data set comprises N groups of overlapped internal flaw image sets, and N is a positive integer;
Obtaining a fiber layer defect feature set, and screening the apparent defect data set based on the fiber layer defect feature set to obtain an abnormal defect data set and fiber layer defect information, wherein the abnormal defect data set comprises overlapped defect information and non-identification defect information, and comprises the following steps:
performing flaw feature extraction based on the sample fiber layer flaw image set to obtain the fiber layer flaw feature set;
extracting and obtaining a first overlapped apparent flaw image set based on the H groups of overlapped apparent flaw image sets;
judging whether the first overlapped apparent flaw image set meets the fiber layer flaw feature set or not;
if the first overlapped apparent flaw image set meets the fiber layer flaw feature set, extracting a feature image of the first overlapped apparent flaw image set to obtain a first fiber layer flaw defect;
adding the first fiber layer flaw defect into the fiber layer flaw defect information;
if part of apparent flaw images in the first overlapped apparent flaw image set meet the fiber layer flaw feature set, multi-feature image extraction is carried out on the first overlapped apparent flaw image set, and a first overlapped flaw defect is obtained;
Adding the first overlapped flaw defect to the overlapped flaw defect information;
if the first overlapped apparent flaw image set does not meet the fiber layer flaw feature set, extracting a feature image of the first overlapped apparent flaw image set to obtain a first non-identification flaw defect;
adding the first unidentified flaw defect to the unidentified flaw defect information;
and by analogy, comparing the fiber layer flaw feature set with the H groups of overlapping apparent flaw image sets to obtain an abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapping flaw defect information and non-identification flaw defect information;
performing defect authentication of the internal defect data set based on the abnormal defect data set to obtain liner layer defect information, including:
performing flaw feature extraction on the flaw image set of the sample liner layer to obtain a flaw feature set of the liner layer;
performing flaw authentication of the non-identification flaw defect information based on the flaw feature set of the liner layer to obtain a flaw defect set of the first liner layer;
screening the overlapped flaw defect information according to the flaw feature set of the liner layer and the flaw feature set of the fiber layer to obtain a flaw defect set of the second liner layer and flaw defect compensation of the fiber layer;
Performing single characteristic identification on the N groups of overlapped internal flaw image sets through the flaw characteristic sets of the inner liner layer to obtain a flaw defect set of a third inner liner layer;
performing spatial position aggregation on the first liner layer flaw defect set, the second liner layer flaw defect set and the third liner layer flaw defect set to obtain liner layer flaw defect information;
performing data optimization of the fiber layer flaw defect information based on the fiber layer flaw defect compensation;
and the fiber layer flaw defect information and the liner layer flaw defect information form a flaw defect identification result of the target pressure container.
2. The method of claim 1, wherein the method further comprises:
the fiber layer defect information comprises M fiber layer defect defects, wherein each fiber layer defect has the spatial position mark;
performing region division on the apparent detection surface based on the spatial position mark to obtain an apparent detection region division result;
carrying out hierarchical light intensity irradiation on the apparent detection surface, and adopting the image acquisition device to acquire multi-view images so as to obtain a fiber layer light intensity image set;
Mapping the apparent detection region division result to the fiber layer light intensity image set for division to obtain an apparent detection region image set;
performing flaw size extraction based on the apparent detection area image set to obtain M groups of fiber layer flaw size parameters;
and carrying out serialization treatment on M groups of fiber layer flaw size parameters to obtain M pieces of fiber layer flaw size information.
3. The method of claim 2, wherein M sets of fiber layer flaw size parameters are serialized to obtain M fiber layer flaw size information, and wherein the method further comprises:
the method comprises the steps of interactively obtaining a flaw size constraint set of the fiber layer flaw feature set, wherein the flaw size constraint set and the fiber layer flaw feature set have a mapping relation;
performing defect type identification of the M fiber layer defect defects based on the fiber layer defect feature set to obtain M fiber layer defect type identifiers;
mapping and extracting the flaw size constraint set based on the M fiber layer flaw type identifiers to obtain M flaw size constraints;
judging whether the M fiber layer flaw size information completely meets the M flaw size constraints;
If the M flaw size information of the fiber layers completely meets the M flaw size constraints, generating a fiber layer flaw unobstructed instruction;
and if any one of the M fiber layer flaw size information does not meet the M flaw size constraints, generating a pressure vessel scrapping instruction.
4. A non-destructive inspection system for a nonmetallic liner filament wound container, the system comprising:
the detection surface dividing module is used for dividing the detection surface of the target pressure vessel to obtain an apparent detection surface and an internal detection surface;
the apparent light wave image module is used for carrying out hierarchical light wave irradiation on the apparent detection surface, carrying out multi-view image acquisition by adopting an image acquisition device, and obtaining an apparent hierarchical light wave image set, wherein the apparent hierarchical light wave image set comprises K spliced apparent hierarchical light wave images;
the internal light wave image module is used for carrying out hierarchical light wave irradiation on the internal detection surface, carrying out multi-view image acquisition by adopting the image acquisition device, and obtaining an internal hierarchical light wave image set, wherein the internal hierarchical light wave image set comprises K spliced internal hierarchical light wave images;
The flaw identification acquisition module is used for respectively inputting the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in a container flaw identification model to obtain an apparent flaw identification set and an internal flaw identification set, wherein flaw identifications have space position marks;
the flaw feature aggregation module is used for conducting flaw feature aggregation on the apparent flaw identification set and the internal flaw identification set based on the spatial position marks to obtain an apparent flaw data set and an internal flaw data set;
the defect data screening module is used for obtaining a fiber layer defect feature set, screening the apparent defect data set based on the fiber layer defect feature set and obtaining an abnormal defect data set and fiber layer defect information, wherein the abnormal defect data set comprises overlapped defect information and non-identification defect information;
the internal flaw authentication module is used for carrying out flaw authentication on the internal flaw data set based on the abnormal flaw data set to obtain flaw defect information of the liner layer;
The defect identification result module is used for forming a defect and defect identification result of the target pressure container by the defect and defect information of the fiber layer and the defect and defect information of the liner layer;
the flaw identification acquisition module comprises the following execution steps:
interactively obtaining a target model parameter of the target pressure container;
extracting and obtaining historical flaw detection information based on the target model parameters, wherein the historical flaw detection information comprises a sample fiber layer flaw image set and a sample liner layer flaw image set;
dividing the flaw images and the background images of the flaw image set of the sample fiber layer and the flaw image set of the sample liner layer based on semantic division to obtain a flaw image set of a training sample;
performing synchronous construction of the first recognition module and the second recognition module based on the training sample flaw image set to generate the container flaw recognition model;
respectively inputting the apparent level light wave image set and the internal level light wave image set into a first identification module and a second identification module in the container flaw identification model to obtain the apparent flaw identification set and the internal flaw identification set;
The flaw feature aggregation module comprises the following execution steps:
performing flaw space positioning on the apparent flaw identification set based on the space position mark to obtain the apparent flaw data set, wherein the apparent flaw data set comprises H groups of overlapped apparent flaw image sets, and H is a positive integer;
performing flaw space positioning on the internal flaw identification set based on the space position mark to obtain the internal flaw data set, wherein the internal flaw data set comprises N groups of overlapped internal flaw image sets, and N is a positive integer;
the flaw data screening module comprises the following execution steps:
performing flaw feature extraction based on the sample fiber layer flaw image set to obtain the fiber layer flaw feature set;
extracting and obtaining a first overlapped apparent flaw image set based on the H groups of overlapped apparent flaw image sets;
judging whether the first overlapped apparent flaw image set meets the fiber layer flaw feature set or not;
if the first overlapped apparent flaw image set meets the fiber layer flaw feature set, extracting a feature image of the first overlapped apparent flaw image set to obtain a first fiber layer flaw defect;
Adding the first fiber layer flaw defect into the fiber layer flaw defect information;
if part of apparent flaw images in the first overlapped apparent flaw image set meet the fiber layer flaw feature set, multi-feature image extraction is carried out on the first overlapped apparent flaw image set, and a first overlapped flaw defect is obtained;
adding the first overlapped flaw defect to the overlapped flaw defect information;
if the first overlapped apparent flaw image set does not meet the fiber layer flaw feature set, extracting a feature image of the first overlapped apparent flaw image set to obtain a first non-identification flaw defect;
adding the first unidentified flaw defect to the unidentified flaw defect information;
and by analogy, comparing the fiber layer flaw feature set with the H groups of overlapping apparent flaw image sets to obtain an abnormal flaw data set and fiber layer flaw defect information, wherein the abnormal flaw data set comprises overlapping flaw defect information and non-identification flaw defect information;
the internal flaw authentication module comprises the following execution steps:
performing flaw feature extraction on the flaw image set of the sample liner layer to obtain a flaw feature set of the liner layer;
Performing flaw authentication of the non-identification flaw defect information based on the flaw feature set of the liner layer to obtain a flaw defect set of the first liner layer;
screening the overlapped flaw defect information according to the flaw feature set of the liner layer and the flaw feature set of the fiber layer to obtain a flaw defect set of the second liner layer and flaw defect compensation of the fiber layer;
performing single characteristic identification on the N groups of overlapped internal flaw image sets through the flaw characteristic sets of the inner liner layer to obtain a flaw defect set of a third inner liner layer;
performing spatial position aggregation on the first liner layer flaw defect set, the second liner layer flaw defect set and the third liner layer flaw defect set to obtain liner layer flaw defect information;
and carrying out data optimization on the defect information of the fiber layer defects based on the defect compensation of the fiber layer defects.
CN202310726873.XA 2023-06-19 2023-06-19 Nondestructive testing method for nonmetallic liner fiber winding container Active CN116754484B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310726873.XA CN116754484B (en) 2023-06-19 2023-06-19 Nondestructive testing method for nonmetallic liner fiber winding container

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310726873.XA CN116754484B (en) 2023-06-19 2023-06-19 Nondestructive testing method for nonmetallic liner fiber winding container

Publications (2)

Publication Number Publication Date
CN116754484A CN116754484A (en) 2023-09-15
CN116754484B true CN116754484B (en) 2024-01-05

Family

ID=87952899

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310726873.XA Active CN116754484B (en) 2023-06-19 2023-06-19 Nondestructive testing method for nonmetallic liner fiber winding container

Country Status (1)

Country Link
CN (1) CN116754484B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116993733B (en) * 2023-09-27 2024-01-09 深圳市冉希古科技有限公司 Earphone sleeve appearance quality detection method and system
CN117152153B (en) * 2023-10-31 2024-01-26 南通顺裕包装材料有限公司 Plastic packaging bottle body flaw detection method and system
CN117871415A (en) * 2024-03-11 2024-04-12 天津大学四川创新研究院 Exposure type structural flaw detection system and method based on parallel light source

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111709948A (en) * 2020-08-19 2020-09-25 深兰人工智能芯片研究院(江苏)有限公司 Method and device for detecting defects of container
CN113592828A (en) * 2021-08-03 2021-11-02 南京市特种设备安全监督检验研究院 Nondestructive testing method and system based on industrial endoscope
CN114743102A (en) * 2022-04-11 2022-07-12 中山大学 Furniture board oriented flaw detection method, system and device
CN116008289A (en) * 2023-02-07 2023-04-25 浙江聚优非织造材料科技有限公司 Nonwoven product surface defect detection method and system

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111709948A (en) * 2020-08-19 2020-09-25 深兰人工智能芯片研究院(江苏)有限公司 Method and device for detecting defects of container
CN113592828A (en) * 2021-08-03 2021-11-02 南京市特种设备安全监督检验研究院 Nondestructive testing method and system based on industrial endoscope
CN114743102A (en) * 2022-04-11 2022-07-12 中山大学 Furniture board oriented flaw detection method, system and device
CN116008289A (en) * 2023-02-07 2023-04-25 浙江聚优非织造材料科技有限公司 Nonwoven product surface defect detection method and system

Also Published As

Publication number Publication date
CN116754484A (en) 2023-09-15

Similar Documents

Publication Publication Date Title
CN116754484B (en) Nondestructive testing method for nonmetallic liner fiber winding container
CN107771212B (en) Colony contrast collection
JP4908995B2 (en) Defect classification method and apparatus, and defect inspection apparatus
WO2018061067A1 (en) Defect inspection device and defect inspection method
JP4213142B2 (en) Colony counting method
JP5342606B2 (en) Defect classification method and apparatus
CN111398291B (en) Flat enameled electromagnetic wire surface flaw detection method based on deep learning
CN111985292A (en) Microscopy method for image processing results, microscope and computer program with verification algorithm
CN116678826A (en) Appearance defect detection system and method based on rapid three-dimensional reconstruction
CN116596889A (en) X-ray industrial detection defect identification method and system based on deep learning
CN216525503U (en) Carbon fiber prepreg surface defect on-line measuring device based on machine vision
CN112703531A (en) Generating annotation data for tissue images
CN113853607A (en) System and method for monitoring bacterial growth and predicting colony biomass of colonies
CN117198909A (en) Wafer defect detection device and method and readable storage medium
JP5530126B2 (en) Three-dimensional cell image analysis system and three-dimensional cell image analyzer used therefor
JPH10185911A (en) Device for analyzing cell and method therefor
KR101782366B1 (en) Vision inspection method based on learning data using normalization of sample scale
CN113178228B (en) Cell analysis method based on nuclear DNA analysis, computer device, and storage medium
JP2014157158A (en) Cell observation method, three-dimensional cell image analysis system, and three-dimensional cell image analyzer used therefor
CN114926675A (en) Method and device for detecting shell stain defect, computer equipment and storage medium
CN117853826B (en) Object surface precision identification method based on machine vision and related equipment
CN116612075A (en) Defect detection method, defect detection device, electron beam measurement equipment and computer storage medium
WO2021046759A1 (en) Apparatus, system and method for quality control of polymer pellets and method for controlling production line for producing polymer pellets
CN117542301A (en) Display screen detection method and system
CN117872699A (en) Watch sealing rubber ring defect detection system, method and device

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
GR01 Patent grant
GR01 Patent grant