CN117351062A - Fan blade defect diagnosis method, device and system and electronic equipment - Google Patents
Fan blade defect diagnosis method, device and system and electronic equipment Download PDFInfo
- Publication number
- CN117351062A CN117351062A CN202311646560.XA CN202311646560A CN117351062A CN 117351062 A CN117351062 A CN 117351062A CN 202311646560 A CN202311646560 A CN 202311646560A CN 117351062 A CN117351062 A CN 117351062A
- Authority
- CN
- China
- Prior art keywords
- fan blade
- image
- images
- defect
- rotated
- 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.)
- Granted
Links
- 230000007547 defect Effects 0.000 title claims abstract description 199
- 238000003745 diagnosis Methods 0.000 title claims abstract description 114
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000012545 processing Methods 0.000 claims abstract description 20
- 238000001514 detection method Methods 0.000 claims description 97
- 238000005070 sampling Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 description 16
- 238000010586 diagram Methods 0.000 description 14
- 238000005516 engineering process Methods 0.000 description 13
- 230000007246 mechanism Effects 0.000 description 12
- 230000006870 function Effects 0.000 description 8
- 238000012360 testing method Methods 0.000 description 8
- 238000012795 verification Methods 0.000 description 6
- 230000000007 visual effect Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 238000005299 abrasion Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000007797 corrosion Effects 0.000 description 3
- 238000005260 corrosion Methods 0.000 description 3
- 239000004576 sand Substances 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000012774 diagnostic algorithm Methods 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000011045 prefiltration Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
- G01N21/9515—Objects of complex shape, e.g. examined with use of a surface follower device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/70—Wind energy
- Y02E10/72—Wind turbines with rotation axis in wind direction
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- Multimedia (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a fan blade defect diagnosis method, device and system and electronic equipment. The method comprises the following steps: obtaining fan blade images of a plurality of positions of fan blades to obtain a plurality of fan blade images; processing each fan blade image respectively to obtain a fan blade angle corresponding to each fan blade image; rotating the fan blade images according to the corresponding fan blade angles aiming at each fan blade image so as to enable the fan blades in the fan blade images to be at preset angles, obtaining rotated images, performing defect diagnosis on the rotated images, and obtaining a first defect area in the rotated images; and obtaining a diagnosis result of the fan blade according to the first defect area in each rotated image.
Description
Technical Field
The invention relates to the technical field of fans, in particular to a fan blade defect diagnosis method, device and system and electronic equipment.
Background
The fan diagnosis method in the related art is based on the traditional target detection algorithm, and when the fan is diagnosed, because the target detection frames of the traditional target detection marking data and the diagnosis data are positive rectangular frames, the fan blade angle is not fixed and cannot be adapted, and therefore accurate diagnosis of defects cannot be realized. Moreover, the diagnosis technology in the related art can only perform defect fault diagnosis, and cannot accurately diagnose the specific position of the defect fault on the fan blade, and maintenance personnel are required to manually locate the position of the defect fault, so that trouble is brought to the maintenance personnel, and the maintenance efficiency cannot be improved.
Disclosure of Invention
The present invention aims to solve at least one of the technical problems in the related art to some extent. Therefore, a first object of the present invention is to provide a fan blade defect diagnosis method to realize accurate diagnosis of fan blades.
A second object of the invention is to propose an electronic device.
A third objective of the present invention is to provide a fan blade defect diagnosis device.
The fourth objective of the present invention is to provide a fan blade defect diagnosis system.
In order to achieve the above objective, an embodiment of a first aspect of the present invention provides a method for diagnosing a fan blade defect, the method comprising: obtaining fan blade images of a plurality of positions of fan blades to obtain a plurality of fan blade images; processing each fan blade image to obtain a fan blade angle corresponding to each fan blade image; rotating the fan blade images according to the corresponding fan blade angles aiming at each fan blade image so as to enable the fan blades in the fan blade images to be at preset angles, obtaining rotated images, and performing defect diagnosis on the rotated images to obtain a first defect area in the rotated images; and obtaining a diagnosis result of the fan blade according to the first defect area in each rotated image.
In order to achieve the above object, an embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, where the computer program is executed by the processor to implement the method for diagnosing fan blade defects.
To achieve the above object, an embodiment of a third aspect of the present invention provides a fan blade defect diagnosis device, including: the acquisition module is used for acquiring fan blade images of a plurality of positions of fan blades to obtain a plurality of fan blade images; the processing module is used for respectively processing each fan blade image to obtain a fan blade angle corresponding to each fan blade image; the rotating module is used for rotating the fan blade images according to the corresponding fan blade angles aiming at each fan blade image so as to enable the fan blades in the fan blade images to be at preset angles and obtain rotated images; the diagnosis module is used for carrying out defect diagnosis on the rotated images to obtain first defect areas in the rotated images, and obtaining diagnosis results of the fan blades according to the first defect areas in the rotated images.
In order to achieve the above objective, a fourth embodiment of the present invention provides a fan blade defect diagnosis system, which includes the fan blade defect diagnosis device.
According to the fan blade defect diagnosis method, device and system and electronic equipment, fan blade images of a plurality of positions of a fan blade are acquired, a plurality of fan blade images are obtained, each fan blade image is processed to obtain a fan blade angle corresponding to each fan blade image, the fan blade image is rotated according to the corresponding fan blade angle for each fan blade image, so that the fan blades in the fan blade image are at a preset angle, a rotated image is obtained, defect diagnosis is conducted on the rotated image, a first defect area in the rotated image is obtained, and a diagnosis result of the fan blade is obtained according to the first defect area in each rotated image. The fan blade image is rotated at first, then the fan blade image is detected, the problem that a target detection method in the related technology cannot detect targets which are arranged in a non-horizontal and non-vertical mode can be avoided, the problem that the rotating target detection in the related technology can detect the targets which are arranged in the non-horizontal and non-vertical mode, but accuracy is lost, angle loss can be caused, because of the angle boundary problem, the problem that a loss value increases suddenly when the fan blade is positioned at the boundary angle is solved, the fan blade is detected with high accuracy, the visual detection frame for detecting the rotating target can be achieved through detection in the method, and meanwhile the accuracy of traditional target detection is also achieved. And moreover, fan blade images at a plurality of positions are obtained by shooting fan blades, each fan blade image is detected, and the detected fan blade images are combined, so that the specific defect position can be determined by the fan blade image with the detected defect, and the accurate judgment of the defect position is realized.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a flow chart of a method of diagnosing fan blade defects in accordance with one or more embodiments of the present invention;
FIG. 2 is a graph showing a rotation result of a fan blade image according to an example of the present invention;
FIG. 3 is a flow chart of a method of diagnosing fan blade defects in accordance with one or more embodiments of the present invention;
FIG. 4 is a fan blade gray scale map of an example of the present invention;
FIG. 5 is a fan blade pixel profile for one example of the present invention;
FIG. 6 is a diagram of fan blade pixel profiles for another example of the present invention;
FIG. 7 is a schematic illustration of a fan blade image anti-rotation result according to an example of the present invention;
FIG. 8 is a schematic view of a fan blade image stitching according to an example of the present invention;
FIG. 9 is a schematic diagram illustrating the results of a fan blade defect diagnosis method according to an example of the present invention;
FIG. 10 is a schematic view of an exemplary fan blade of the present invention;
FIG. 11 is a schematic diagram of an exemplary object detection algorithm;
FIG. 12 is a sample point schematic of an example convolution operation;
fig. 13 (a), 13 (b), 13 (c) are schematic diagrams of sampling points according to an example of the present invention;
FIG. 14 is a schematic diagram of an exemplary fan blade defect;
FIG. 15 is a schematic view of another exemplary fan blade defect;
FIG. 16 is a schematic diagram of the operation of a variable convolution of one example of the present invention;
FIG. 17 is a schematic diagram of an exemplary attention mechanism of the present invention;
FIG. 18 is a flow chart of a fan blade defect diagnosis method according to an example of the present invention;
FIG. 19 is a block diagram illustrating a fan blade defect diagnosis apparatus according to an embodiment of the present invention;
FIG. 20 is a block diagram of a fan blade defect diagnosis system according to an embodiment of the present invention.
Detailed Description
The method, apparatus, system and electronic device for diagnosing defects of fan blades according to embodiments of the present invention are described below with reference to the accompanying drawings, in which the same or similar reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described with reference to the drawings are exemplary and should not be construed as limiting the invention.
FIG. 1 is a flow chart of a method of diagnosing fan blade defects in accordance with one or more embodiments of the present invention.
As shown in fig. 1, the fan blade defect diagnosis method includes:
s11, obtaining fan blade images of a plurality of positions of fan blades, and obtaining a plurality of fan blade images.
S12, processing each fan blade image to obtain a fan blade angle corresponding to each fan blade image.
S13, rotating the fan blade images according to the corresponding fan blade angles aiming at each fan blade image so as to enable the fan blades in the fan blade images to be at a preset angle, obtaining rotated images, and performing defect diagnosis on the rotated images to obtain a first defect area in the rotated images.
S14, obtaining a diagnosis result of the fan blade according to the first defect area in each rotated image.
Specifically, first, a plurality of positions of fan blades are photographed, and a plurality of blade images of the fan blades are obtained. Because a plurality of positions of the fan blade are shot, the fan blade can be shot at a short distance, so that the fan blade is displayed more finely on a shot image, and the diagnosis capability of the fan blade is improved.
After the fan blade image is obtained, the angle of the fan blade in the obtained fan blade image is obtained, and the fan blade image is rotated according to the angle of the fan blade. And then carrying out defect detection on the rotated fan blade image to obtain a defect area.
The structure of the target detection network determines that the target detection frames of the target detection labeling data and the diagnostic data are positive rectangular frames, so that when the fan blade image is rotated according to the fan blade angle, the fan blades in the image are required to be horizontal or vertical after rotating, and the target detection frames can be directly adopted to realize the addition of the rotating target detection frames. The specific rotation of the fan blade image may be seen in the example shown in fig. 2. After the fan blade image is rotated, a positive rectangular frame is used for the fan blade image.
The method comprises the steps of obtaining fan blade images of a plurality of positions of a fan blade, respectively processing each fan blade image to obtain fan blade angles corresponding to each fan blade image, rotating the fan blade images according to the corresponding fan blade angles for each fan blade image so that the fan blades in the fan blade images are at preset angles to obtain rotated images, performing defect diagnosis on the rotated images to obtain first defect areas in the rotated images, and obtaining a diagnosis result of the fan blade according to the first defect areas in each rotated image. The fan blade image is rotated at first, then the fan blade image is detected, the problem that a target detection method in the related technology cannot detect targets which are arranged in a non-horizontal and non-vertical mode can be avoided, the problem that the rotating target detection in the related technology can detect the targets which are arranged in the non-horizontal and non-vertical mode, but accuracy is lost, angle loss can be caused, because of the angle boundary problem, the problem that a loss value increases suddenly when the fan blade is positioned at the boundary angle is solved, the fan blade is detected with high accuracy, the visual detection frame for detecting the rotating target can be achieved through detection in the method, and meanwhile the accuracy of traditional target detection is also achieved. And moreover, fan blade images at a plurality of positions are obtained by shooting fan blades, each fan blade image is detected, and the detected fan blade images are combined, so that the specific defect position can be determined by detecting the defective fan blade image, and the accurate judgment of the defect position is realized. Through the target detection frame, most of background negative samples caused by the fact that the whole frame selects the blades are avoided, and position information of defects is marked more intuitively.
In one or more embodiments of the present invention, before obtaining the diagnosis result of the fan blade according to the first defect area in each rotated image, referring to fig. 3, the fan blade defect diagnosis method further includes:
s31, converting the fan blade image into a gray scale image.
S32, carrying out gray value distribution statistics on the gray level map to obtain gray value distribution of the gray level map.
S33, obtaining a second defect area in the gray scale map according to the gray scale value distribution.
Specifically, before sampling, the image is first processed to generate a gray map using the gray function of opencv and the calculated fan blade angle rotation function. In addition, when the gray level map is generated, the fan blade image is required to be segmented, so that interference of background redundant information is reduced, the time of network training is shortened, and time expenditure is reduced to improve accuracy. Taking the fan blade shown in fig. 2 as an example, the generated gray-scale map may be as shown in fig. 4.
For the front edge of the fan blade with erosion and damage conditions such as sand blast abrasion, when the whole fan blade is used for carrying out pixel distribution statistics, besides a little obvious fluctuation of the background and the foreground, the fluctuation of the pixel value of the fan blade needs to be considered. The fluctuation value of the fan blade itself can be seen in the examples shown in fig. 5 and 6. Fig. 5 and 6 are images obtained by plotting by matplotlib, and the foreground of the defect detection is segmented from fig. 4, so as to perform distribution statistics of pixel values. Fig. 5 employs pli.hist () function. Fig. 6 uses the plt.plot (cv.calchist ()) function to count the number of pixels per pixel for the statistical distribution of image pixels. Fig. 5 shows a pixel distribution, in which the abscissa indicates pixel values 0 to 255 and the ordinate indicates a total of 256 subsets (0 to 255) of the corresponding pixel values. Fig. 6 is a diagram showing the number of each gradation value in the statistical image.
The normal fan blade without corrosion has the gray image pixel values uniformly distributed, and the pixel values are uniformly distributed between 0 and 80 in a concentrated manner when interference of other factors such as light rays is eliminated. Therefore, after the gray level image is obtained, gray level value distribution statistics can be carried out on the gray level image, gray level value distribution of the gray level image is obtained, and further, after the first defect area is obtained, verification is carried out again to determine whether fan blades are corroded or not through comparison of the difference values of the wave peaks which are distributed intensively.
Further, after the verification is completed, a diagnosis result of the fan blade can be obtained according to the first defect area and the second defect area. Specifically, a proportion weight may be set, for example, the proportion of the defect diagnosis is 0.6, the proportion of the corrosion diagnosis is 0.4, after the defect diagnosis result and the corrosion diagnosis result are obtained, the proportion is multiplied, the calculated result is compared with a preset threshold value interval, and whether the fan blade is corroded such as sand and wind abrasion or not is judged according to the comparison result.
As the front edge of the fan almost has the abrasion of wind and sand, the blade tip with the shape is taken as the type to be identified and detected during the characteristic extraction, thereby neglecting the abrasion defect characteristics. Therefore, the fan blade is set to be corroded and verified, so that the situation is avoided.
In one or more embodiments of the present invention, after obtaining a diagnosis result of the fan blade according to the first defect area in each rotated image, the fan blade defect diagnosis method further includes: respectively reversely rotating each first defect area to an angle before rotation along with the corresponding rotated image to obtain a reversely rotated image; and splicing all the anti-rotated images into a complete image, and displaying the first defect area after the anti-rotation in the complete image.
Specifically, after the diagnosis result is obtained, the image of each fan blade is subjected to anti-rotation operation, for example, the image is rotated clockwise by 8 degrees in the rotation operation, and the anti-rotation operation is rotated anticlockwise by 8 degrees, so that the image is restored. Taking the rotated fan image shown in fig. 2 as an example, the counter rotated fan image may be as shown in fig. 7.
After each fan blade image is restored, carrying out algorithm diagnosis and splicing on the shot local images, and restoring the original condition of the whole fan blade according to the fan blade angle calculated before inspection.
The unmanned aerial vehicle can fly to take photos along the blade direction by setting the navigation point information preset according to the field angle of the camera and the shooting setting. In the flying photographing process, the unmanned aerial vehicle keeps the same with the direction of the blade, so that the adjacent two pictures can be spliced according to the overlapped pixel value and the angle of the blade. Such as: the X-axis direction is overlapped by 40 pixels, the Y-axis direction is overlapped by 200 pixels, and the position information of each picture is calculated according to the measuring angle. It is then subjected to rotational clipping as shown in fig. 8. The rotation angle is the angle value of the measured fan blade before inspection, a two-dimensional rotation transformation matrix such as getrotation matrix2D (a function of obtaining a rotation matrix of an image around a certain point) is used, and an effect diagram finally shown in fig. 9 is generated and displayed on a terminal such as a remote controller. According to the spliced pictures, the positions of the defect faults can be clearly positioned in the fan blades, and great convenience is brought to maintenance personnel.
In one or more embodiments of the present invention, before rotating the fan blade image according to the corresponding fan blade angle for each fan blade image, the fan blade defect diagnosis method further includes: and adopting a PnP sampling module to carry out foreground extraction processing on the fan blade image so as to carry out sharpening processing on the fan blade in the fan blade image.
Specifically, in the stage of extracting the training defect feature of the network model, a PnP sampling module is adopted to reduce redundant calculation of a background area, and meanwhile, image feature mapping is abstracted into a fine foreground object feature vector and a small amount of rough background context feature vector as shown in fig. 10. The feature map may be considered as an "abstraction" of the input data in the convolutional neural network, which may extract information of different features in the input data, such as edges, textures, shapes, etc. As the depth of the network increases, feature maps become more abstract, and can extract higher-level features, such as part, whole, category and other information of objects, and the feature maps can be finally sent into a classifier for classification or regression and other tasks through pooling, full connection and other operations.
In one or more embodiments of the present invention, before the fan blade image is subjected to the foreground extraction processing by using the PnP sampling module, the fan blade defect diagnosis method further includes: and carrying out convolution processing on the fan blade image by adopting a deformable convolution network so as to realize small target detection on the fan blade image.
The complex variety of different defects and the large difference of the similar defects bring great difficulty to detection by adopting a target detection algorithm of an attention adding mechanism as shown in fig. 11. The variety of different defects is complex and mainly represented in three aspects. Firstly, the difference between the types is large, the appearance defects of fan blades are complex and various, the morphological characteristics of the defects of different types are possibly extremely large, the universality of a detection algorithm is not strong due to the difference, and a plurality of defects need to develop the detection algorithm independently, so that the development complexity is extremely high. And secondly, the inter-class ambiguity is large, and the inter-class ambiguity is the other extreme with large inter-class difference, namely, the apparent features of defects in different classes have certain similarity, the types of the defects are difficult to distinguish, and the products cannot be accurately graded. Third, the background is complex, and it is difficult to completely separate the defect and the background in the production scene, and the defect features are not obvious. This presents a significant challenge to diagnostic algorithms, which have not been able to meet the accuracy requirements. Based on the reality condition, the improved diagnosis algorithm DETR (target detection algorithm) is adopted, so that accuracy can be improved, model training time can be greatly shortened, and convergence speed can be greatly improved.
DETR also has its own problems: 1. he needs longer training time to converge than existing detectors, on coco dataset (a dataset that can be used for image detection, semantic segmentation and image header generation), he needs 500 rounds to converge, 10 to 20 times that of the master r-cnn (a target detection network); DETR performs poorly on small object detection, existing detectors typically have multi-scale features, small object targets are typically detected on high-resolution feature maps, whereas DETR does not detect with multi-scale features, mainly high-resolution feature maps add unacceptable computational complexity to DETR and thus have poor detection of small objects. Here, a formable DETR: the deformable convolution DCN is added on the basis of the DETR, and the deformable convolution is a powerful and efficient mechanism for processing the sparse space position, so that the problem of small target detection is naturally avoided. The deformable convolution allows for a small set of sample locations at all feature image pixels to be a pre-filter salient key element; the module can be naturally extended to aggregate multi-scale features without the help of FPN (Feature Pyramid Networks, feature pyramid network).
The sampling points of the normal convolution operation and the deformable convolution can be as shown in fig. 12 and fig. 13 (a), fig. 13 (b), and fig. 13 (c). Fig. 12 is a normal convolution operation, and fig. 13 (a), 13 (b), and 13 (c) are deformable convolutions. The deformable convolution is subjected to deformation operation, and because the detected defects are complex and various in appearance, different defect detection needs to be dealt with by designing a unique deformable convolution mode, so that compatibility and matching are higher, and the model fitting effect is better. Defects such as a breakage defect indicated by an arrow in fig. 14, a scratch defect indicated by an arrow in fig. 15, and the like can be detected.
The deformable convolution adds an offset to the sampling position so that the convolution kernel can be extended to a large extent during training. Fig. 13 (b) and (c) are special cases of fig. 13 (a), showing that the deformable convolution generalizes various variations in scale, aspect ratio and rotation. Therefore, the irregular shape of the object can be matched more accurately, the accuracy of image processing is improved, and compatibility is brought to large defect target detection. As shown in fig. 16.
Most modern object detection frameworks benefit from multi-scale feature maps. The deformable attention module may naturally be expanded into a multi-scale feature map. The attention mechanism employed in this embodiment can be seen in fig. 17. The multiscale deformable attention module is applied as follows:
Where M represents the attention head. k represents a sampling point. K represents the total number of sampling points.Representing the attention weights of the kth sample point in the ith feature layer and mth attention header. L is the total number of feature layers. Attention weight->The value range of (2) is [0,1 ]]By->Normalization was performed. />Is a 2-d real number with unlimited range. Normalized coordinates are adopted->To represent the sharpness of the MSDeformAttn () formula, with normalized coordinates (0, 0) and (1, 1) representing the upper left and lower right corners of the image, respectively. Function->Normalized coordinates->Rescaling to the input feature map of the first layer. Multiscale deformable attention is very similar to previous single-scale versions, except that multiscale deformableAttention is sampled from the multi-scale feature map rather than from the single-scale feature map. />To represent the sample offset of the kth sample point in the ith feature layer and mth attention header. Wm is the output result of linearly transforming the attention application result to obtain different attention heads. Wm' is a linear map. />Is a scalar representing the attention value derived from the mth attention head and the kth sample point. And x is an input feature map. P is p qx As reference point p q Is p qy As reference point p q Is defined by the vertical coordinate of (c).
In one or more embodiments of the present invention, a specific fan blade defect diagnosis method may be referred to as an example shown in fig. 18.
In summary, in the fan blade defect diagnosis method of the embodiment of the invention, fan blade images of a plurality of positions of a fan blade are acquired, a plurality of fan blade images are obtained, each fan blade image is processed to obtain a fan blade angle corresponding to each fan blade image, the fan blade image is rotated according to the corresponding fan blade angle for each fan blade image, so that the fan blade in the fan blade image is at a preset angle to obtain a rotated image, defect diagnosis is performed on the rotated image to obtain a first defect area in the rotated image, and a diagnosis result of the fan blade is obtained according to the first defect area in each rotated image. The fan blade image is rotated at first, then the fan blade image is detected, the problem that a target detection method in the related technology cannot detect targets which are arranged in a non-horizontal and non-vertical mode can be avoided, the problem that the rotating target detection in the related technology can detect the targets which are arranged in the non-horizontal and non-vertical mode, but accuracy is lost, angle loss can be caused, because of the angle boundary problem, the problem that a loss value increases suddenly when the fan blade is positioned at the boundary angle is solved, the fan blade is detected with high accuracy, the visual detection frame for detecting the rotating target can be achieved through detection in the method, and meanwhile the accuracy of traditional target detection is also achieved. The defect position information is more intuitively marked because the whole frame selects the blade to cause most of background negative samples. And moreover, fan blade images at a plurality of positions are obtained by shooting fan blades, each fan blade image is detected, and the detected fan blade images are combined, so that the specific defect position can be determined by the fan blade image with the detected defect, and the accurate judgment of the defect position is realized. Moreover, the object detection algorithm in the related art diagnoses the problems of missing detection of small object defect diagnosis, false detection of similar defects and false detection of fan blade defects in an inclined state. In order to solve the problems, the method of improved attention mechanism, rotation target detection and the like is adopted, and the defect diagnosis precision is greatly improved. By testing one hundred fault defects, the traditional target detection algorithm such as YOLOV5, fast-RCNN and the like can cause misjudgment and missing detection of small target defects, the preliminary statistics shows that the misjudgment and missing detection reach 20%, but after an improved attention mechanism is added, the misjudgment and missing detection rate is reduced to 8%. Moreover, the method and the device not only set up the rotation and the anti-rotation of the image, but also set up the verification through the gray level, avoid the problem of losing the detection precision of the traditional rotation target, and solve the problems of redundant calculation of the background area and difficult extraction of the defect characteristics. Through the test of one hundred pictures, compared with the traditional diagnosis method, the diagnosis precision is improved by about twenty percent, and the defect area can be marked and displayed more intuitively.
Further, the invention provides electronic equipment.
In the embodiment of the invention, the electronic equipment comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, and the fan blade defect diagnosis method is realized when the computer program is executed by the processor.
According to the electronic equipment provided by the embodiment of the invention, through realizing the fan blade defect diagnosis method, fan blade images of a plurality of positions of a fan blade are acquired, a plurality of fan blade images are obtained, each fan blade image is processed respectively to obtain a fan blade angle corresponding to each fan blade image, the fan blade image is rotated according to the corresponding fan blade angle so that the fan blade in the fan blade image is at a preset angle, a rotated image is obtained, defect diagnosis is carried out on the rotated image to obtain a first defect area in the rotated image, and a diagnosis result of the fan blade is obtained according to the first defect area in each rotated image. The fan blade image is rotated at first, then the fan blade image is detected, the problem that a target detection method in the related technology cannot detect targets which are arranged in a non-horizontal and non-vertical mode can be avoided, the problem that the rotating target detection in the related technology can detect the targets which are arranged in the non-horizontal and non-vertical mode, but accuracy is lost, angle loss can be caused, because of the angle boundary problem, the problem that a loss value increases suddenly when the fan blade is positioned at the boundary angle is solved, the fan blade is detected with high accuracy, the visual detection frame for detecting the rotating target can be achieved through detection in the method, and meanwhile the accuracy of traditional target detection is also achieved. The defect position information is more intuitively marked because the whole frame selects the blade to cause most of background negative samples. And moreover, fan blade images at a plurality of positions are obtained by shooting fan blades, each fan blade image is detected, and the detected fan blade images are combined, so that the specific defect position can be determined by the fan blade image with the detected defect, and the accurate judgment of the defect position is realized. Moreover, the object detection algorithm in the related art diagnoses the problems of missing detection of small object defect diagnosis, false detection of similar defects and false detection of fan blade defects in an inclined state. In order to solve the problems, the method of improved attention mechanism, rotation target detection and the like is adopted, and the defect diagnosis precision is greatly improved. By testing one hundred fault defects, the traditional target detection algorithm such as YOLOV5, fast-RCNN and the like can cause misjudgment and missing detection of small target defects, the preliminary statistics shows that the misjudgment and missing detection reach 20%, but after an improved attention mechanism is added, the misjudgment and missing detection rate is reduced to 8%. Moreover, the method and the device not only set up the rotation and the anti-rotation of the image, but also set up the verification through the gray level, avoid the problem of losing the detection precision of the traditional rotation target, and solve the problems of redundant calculation of the background area and difficult extraction of the defect characteristics. Through the test of one hundred pictures, compared with the traditional diagnosis method, the diagnosis precision is improved by about twenty percent, and the defect area can be marked and displayed more intuitively.
Furthermore, the invention provides a fan blade defect diagnosis device.
FIG. 19 is a block diagram illustrating a fan blade defect diagnosis apparatus according to an embodiment of the present invention.
As shown in fig. 19, the fan blade defect diagnosis device 100 includes: the acquisition module 101 is configured to acquire fan blade images of a plurality of positions of a fan blade, and obtain a plurality of fan blade images; the processing module 102 is configured to process each fan blade image respectively to obtain a fan blade angle corresponding to each fan blade image; the rotation module 103 is configured to rotate, for each fan blade image, the fan blade image according to the corresponding fan blade angle, so that the fan blades in the fan blade image are at a preset angle, and a rotated image is obtained; the diagnostic module 104 is configured to perform defect diagnosis on the rotated images to obtain a first defect area in the rotated images, and obtain a diagnostic result of the fan blade according to the first defect area in each rotated image.
In one embodiment of the present invention, the fan blade defect diagnosis device 100 further includes: the conversion module is used for converting the fan blade image into a gray level image; the statistics module is used for carrying out gray value distribution statistics on the gray level images to obtain gray value distribution of the gray level images; the diagnostic module 104 is also configured to: and obtaining a second defect area in the gray level graph according to the gray level distribution, and obtaining a diagnosis result of the fan blade according to the first defect area and the second defect area.
In one embodiment of the invention, the rotation module is further configured to: after the diagnosis module obtains the diagnosis result, respectively reversely rotating each first defect area to an angle before rotation along with the corresponding rotated image to obtain a reversely rotated image; the device also comprises a splicing module and a display module, wherein the splicing module is used for splicing all the images after the anti-rotation into a complete image, and the display module is used for displaying the first defect area after the anti-rotation in the complete image.
It should be noted that, for other specific implementations of the fan blade defect diagnosis device according to the embodiments of the present invention, reference may be made to the fan blade defect diagnosis method of the foregoing embodiments.
According to the fan blade defect diagnosis device, fan blade images of a plurality of positions of a fan blade are acquired, a plurality of fan blade images are obtained, each fan blade image is processed to obtain a fan blade angle corresponding to each fan blade image, the fan blade images are rotated according to the corresponding fan blade angles for each fan blade image, so that the fan blades in the fan blade images are at preset angles, a rotated image is obtained, defect diagnosis is conducted on the rotated image, a first defect area in the rotated image is obtained, and a diagnosis result of the fan blade is obtained according to the first defect area in each rotated image. The fan blade image is rotated at first, then the fan blade image is detected, the problem that a target detection method in the related technology cannot detect targets which are arranged in a non-horizontal and non-vertical mode can be avoided, the problem that the rotating target detection in the related technology can detect the targets which are arranged in the non-horizontal and non-vertical mode, but accuracy is lost, angle loss can be caused, because of the angle boundary problem, the problem that a loss value increases suddenly when the fan blade is positioned at the boundary angle is solved, the fan blade is detected with high accuracy, the visual detection frame for detecting the rotating target can be achieved through detection in the method, and meanwhile the accuracy of traditional target detection is also achieved. The defect position information is more intuitively marked because the whole frame selects the blade to cause most of background negative samples. And moreover, fan blade images at a plurality of positions are obtained by shooting fan blades, each fan blade image is detected, and the detected fan blade images are combined, so that the specific defect position can be determined by the fan blade image with the detected defect, and the accurate judgment of the defect position is realized. Moreover, the object detection algorithm in the related art diagnoses the problems of missing detection of small object defect diagnosis, false detection of similar defects and false detection of fan blade defects in an inclined state. In order to solve the problems, the method of improved attention mechanism, rotation target detection and the like is adopted, and the defect diagnosis precision is greatly improved. By testing one hundred fault defects, the traditional target detection algorithm such as YOLOV5, fast-RCNN and the like can cause misjudgment and missing detection of small target defects, the preliminary statistics shows that the misjudgment and missing detection reach 20%, but after an improved attention mechanism is added, the misjudgment and missing detection rate is reduced to 8%. Moreover, the method and the device not only set up the rotation and the anti-rotation of the image, but also set up the verification through the gray level, avoid the problem of losing the detection precision of the traditional rotation target, and solve the problems of redundant calculation of the background area and difficult extraction of the defect characteristics. Through the test of one hundred pictures, compared with the traditional diagnosis method, the diagnosis precision is improved by about twenty percent, and the defect area can be marked and displayed more intuitively.
Furthermore, the invention provides a fan blade defect diagnosis system.
FIG. 20 is a block diagram of a fan blade defect diagnosis system according to an embodiment of the present invention.
As shown in fig. 20, the fan blade defect diagnosis system 10 includes the fan blade defect diagnosis device 100 described above.
According to the fan blade defect diagnosis system, through the fan blade defect diagnosis device, fan blade images of a plurality of positions of a fan blade are acquired, a plurality of fan blade images are obtained, each fan blade image is processed to obtain a fan blade angle corresponding to each fan blade image, the fan blade image is rotated according to the corresponding fan blade angle for each fan blade image, so that the fan blades in the fan blade image are at a preset angle, a rotated image is obtained, defect diagnosis is conducted on the rotated image, a first defect area in the rotated image is obtained, and a diagnosis result of the fan blade is obtained according to the first defect area in each rotated image. The fan blade image is rotated at first, then the fan blade image is detected, the problem that a target detection method in the related technology cannot detect targets which are arranged in a non-horizontal and non-vertical mode can be avoided, the problem that the rotating target detection in the related technology can detect the targets which are arranged in the non-horizontal and non-vertical mode, but accuracy is lost, angle loss can be caused, because of the angle boundary problem, the problem that a loss value increases suddenly when the fan blade is positioned at the boundary angle is solved, the fan blade is detected with high accuracy, the visual detection frame for detecting the rotating target can be achieved through detection in the method, and meanwhile the accuracy of traditional target detection is also achieved. The defect position information is more intuitively marked because the whole frame selects the blade to cause most of background negative samples. And moreover, fan blade images at a plurality of positions are obtained by shooting fan blades, each fan blade image is detected, and the detected fan blade images are combined, so that the specific defect position can be determined by the fan blade image with the detected defect, and the accurate judgment of the defect position is realized. Moreover, the object detection algorithm in the related art diagnoses the problems of missing detection of small object defect diagnosis, false detection of similar defects and false detection of fan blade defects in an inclined state. In order to solve the problems, the method of improved attention mechanism, rotation target detection and the like is adopted, and the defect diagnosis precision is greatly improved. By testing one hundred fault defects, the traditional target detection algorithm such as YOLOV5, fast-RCNN and the like can cause misjudgment and missing detection of small target defects, the preliminary statistics shows that the misjudgment and missing detection reach 20%, but after an improved attention mechanism is added, the misjudgment and missing detection rate is reduced to 8%. Moreover, the method and the device not only set up the rotation and the anti-rotation of the image, but also set up the verification through the gray level, avoid the problem of losing the detection precision of the traditional rotation target, and solve the problems of redundant calculation of the background area and difficult extraction of the defect characteristics. Through the test of one hundred pictures, compared with the traditional diagnosis method, the diagnosis precision is improved by about twenty percent, and the defect area can be marked and displayed more intuitively.
It should be noted that the logic and/or steps represented in the flow diagrams or otherwise described herein may be considered a ordered listing of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
In the description of the present specification, the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. refer to an orientation or positional relationship based on that shown in the drawings, and do not indicate or imply that the apparatus or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and should not be construed as limiting the invention.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
In the description of the present specification, unless otherwise indicated, the terms "mounted," "connected," "secured," and the like are to be construed broadly and may be, for example, fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; either directly or indirectly, through intermediaries, or both, may be in communication with each other or in interaction with each other, unless expressly defined otherwise. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
In the present invention, unless expressly stated or limited otherwise, a first feature "up" or "down" a second feature may be the first and second features in direct contact, or the first and second features in indirect contact via an intervening medium. Moreover, a first feature being "above," "over" and "on" a second feature may be a first feature being directly above or obliquely above the second feature, or simply indicating that the first feature is level higher than the second feature. The first feature being "under", "below" and "beneath" the second feature may be the first feature being directly under or obliquely below the second feature, or simply indicating that the first feature is less level than the second feature.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.
Claims (10)
1. A fan blade defect diagnosis method is characterized by comprising the following steps:
obtaining fan blade images of a plurality of positions of fan blades to obtain a plurality of fan blade images;
processing each fan blade image to obtain a fan blade angle corresponding to each fan blade image;
Rotating the fan blade images according to the corresponding fan blade angles aiming at each fan blade image so as to enable the fan blades in the fan blade images to be at preset angles, obtaining rotated images, and performing defect diagnosis on the rotated images to obtain a first defect area in the rotated images;
and obtaining a diagnosis result of the fan blade according to the first defect area in each rotated image.
2. The method of claim 1, wherein before obtaining the diagnosis result of the fan blade according to the first defect area in each rotated image, the method further comprises:
converting the fan blade image into a gray scale image;
carrying out gray value distribution statistics on the gray level map to obtain gray value distribution of the gray level map;
obtaining a second defect area in the gray scale map according to the gray scale value distribution;
the obtaining the diagnosis result of the fan blade according to the first defect area in each rotated image includes:
and obtaining a diagnosis result of the fan blade according to the first defect area and the second defect area.
3. The method of claim 1, wherein after obtaining the diagnosis result of the fan blade according to the first defect area in each rotated image, the method further comprises:
Respectively reversely rotating each first defect area to an angle before rotation along with the corresponding rotated image to obtain a reversely rotated image;
and splicing all the anti-rotated images into a complete image, and displaying the first defect area after the anti-rotation in the complete image.
4. The method of claim 1, wherein for each of the fan blade images, before rotating the fan blade image according to the corresponding fan blade angle, the method further comprises:
and carrying out foreground extraction processing on the fan blade image by adopting a PnP sampling module so as to carry out sharpening processing on the fan blade in the fan blade image.
5. The method of claim 4, wherein before performing foreground extraction processing on the fan blade image by using a PnP sampling module, the method further comprises:
and carrying out convolution processing on the fan blade image by adopting a deformable convolution network so as to realize small target detection on the fan blade image.
6. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the fan blade defect diagnosis method of any of claims 1-5.
7. A fan blade defect diagnosis device, the device comprising:
the acquisition module is used for acquiring fan blade images of a plurality of positions of fan blades to obtain a plurality of fan blade images;
the processing module is used for respectively processing each fan blade image to obtain a fan blade angle corresponding to each fan blade image;
the rotating module is used for rotating the fan blade images according to the corresponding fan blade angles aiming at each fan blade image so as to enable the fan blades in the fan blade images to be at preset angles and obtain rotated images;
the diagnosis module is used for carrying out defect diagnosis on the rotated images to obtain first defect areas in the rotated images, and obtaining diagnosis results of the fan blades according to the first defect areas in the rotated images.
8. The fan blade defect diagnosis device of claim 7, further comprising:
the conversion module is used for converting the fan blade image into a gray level image;
the statistics module is used for carrying out gray value distribution statistics on the gray level images to obtain gray value distribution of the gray level images;
the diagnostic module is further configured to:
And obtaining a second defect area in the gray scale map according to the gray scale value distribution, and obtaining a diagnosis result of the fan blade according to the first defect area and the second defect area.
9. The fan blade defect diagnosis device of claim 8, wherein the rotation module is further configured to:
after the diagnosis module obtains the diagnosis result, respectively reversely rotating each first defect area to an angle before rotation along with the corresponding rotated image to obtain a reversely rotated image;
the device also comprises a splicing module and a display module, wherein the splicing module is used for splicing all the images after the anti-rotation into a complete image, and the display module is used for displaying the first defect area after the anti-rotation in the complete image.
10. A fan blade defect diagnosis system, characterized by comprising the fan blade defect diagnosis device according to any one of claims 7 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311646560.XA CN117351062B (en) | 2023-12-04 | 2023-12-04 | Fan blade defect diagnosis method, device and system and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311646560.XA CN117351062B (en) | 2023-12-04 | 2023-12-04 | Fan blade defect diagnosis method, device and system and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117351062A true CN117351062A (en) | 2024-01-05 |
| CN117351062B CN117351062B (en) | 2024-02-23 |
Family
ID=89366974
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311646560.XA Active CN117351062B (en) | 2023-12-04 | 2023-12-04 | Fan blade defect diagnosis method, device and system and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117351062B (en) |
Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102789658A (en) * | 2012-03-02 | 2012-11-21 | 成都三泰电子实业股份有限公司 | Ultraviolet anti-counterfeiting check authenticity verification method |
| CN105488501A (en) * | 2015-11-26 | 2016-04-13 | 南京富士通南大软件技术有限公司 | Method for correcting license plate slant based on rotating projection |
| CN107144569A (en) * | 2017-04-27 | 2017-09-08 | 西安交通大学 | The fan blade surface defect diagnostic method split based on selective search |
| DE102016224988A1 (en) * | 2016-12-14 | 2018-06-14 | Universität Stuttgart | Method and device for detecting defects in an object |
| WO2019104767A1 (en) * | 2017-11-28 | 2019-06-06 | 河海大学常州校区 | Fabric defect detection method based on deep convolutional neural network and visual saliency |
| US20190370999A1 (en) * | 2018-05-31 | 2019-12-05 | Shanghai Clobotics Technology Co., Ltd. | Method, system and equipment for automatically detecting and tracking blade |
| CN111696075A (en) * | 2020-04-30 | 2020-09-22 | 航天图景（北京）科技有限公司 | Intelligent fan blade defect detection method based on double-spectrum image |
| CN112802011A (en) * | 2021-02-25 | 2021-05-14 | 上海电机学院 | Fan blade defect detection method based on VGG-BLS |
| CN113409314A (en) * | 2021-08-18 | 2021-09-17 | 南京市特种设备安全监督检验研究院 | Unmanned aerial vehicle visual detection and evaluation method and system for corrosion of high-altitude steel structure |
| CN113837994A (en) * | 2021-07-29 | 2021-12-24 | 尚特杰电力科技有限公司 | Photovoltaic panel defect diagnosis method based on edge detection convolutional neural network |
| CN114663376A (en) * | 2022-03-15 | 2022-06-24 | 中国华能集团清洁能源技术研究院有限公司 | Fan blade defect detection method and system based on improved SSD model |
| WO2022148109A1 (en) * | 2021-01-05 | 2022-07-14 | 歌尔股份有限公司 | Product defect detection method and apparatus, device and computer-readable storage medium |
| CN115166032A (en) * | 2022-05-23 | 2022-10-11 | 东南大学 | Device and method for detecting cracks of fan blade |
| CN115937114A (en) * | 2022-11-25 | 2023-04-07 | 无锡海纳智能科技有限公司 | Fan picture preprocessing method and device |
| CN116206094A (en) * | 2023-04-28 | 2023-06-02 | 尚特杰电力科技有限公司 | Fan blade angle measuring method, device and system and electronic equipment |
| CN116342607A (en) * | 2023-05-30 | 2023-06-27 | 尚特杰电力科技有限公司 | Power transmission line defect identification method and device, electronic equipment and storage medium |
| EP4227856A1 (en) * | 2022-02-14 | 2023-08-16 | Samsung Display Co., Ltd. | System and method for defect detection |
| CN116797594A (en) * | 2023-07-10 | 2023-09-22 | 华北电力大学（保定） | Fan blade defect detection method, system, electronic equipment and storage medium |
| CN116823872A (en) * | 2023-08-25 | 2023-09-29 | 尚特杰电力科技有限公司 | Fan inspection method and system based on target tracking and image segmentation |
| CN117058396A (en) * | 2023-10-11 | 2023-11-14 | 精效悬浮(苏州)科技有限公司 | Fan blade defect area rapid segmentation method and system based on artificial intelligence |
-
2023
- 2023-12-04 CN CN202311646560.XA patent/CN117351062B/en active Active
Patent Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102789658A (en) * | 2012-03-02 | 2012-11-21 | 成都三泰电子实业股份有限公司 | Ultraviolet anti-counterfeiting check authenticity verification method |
| CN105488501A (en) * | 2015-11-26 | 2016-04-13 | 南京富士通南大软件技术有限公司 | Method for correcting license plate slant based on rotating projection |
| DE102016224988A1 (en) * | 2016-12-14 | 2018-06-14 | Universität Stuttgart | Method and device for detecting defects in an object |
| CN107144569A (en) * | 2017-04-27 | 2017-09-08 | 西安交通大学 | The fan blade surface defect diagnostic method split based on selective search |
| WO2019104767A1 (en) * | 2017-11-28 | 2019-06-06 | 河海大学常州校区 | Fabric defect detection method based on deep convolutional neural network and visual saliency |
| US20190370999A1 (en) * | 2018-05-31 | 2019-12-05 | Shanghai Clobotics Technology Co., Ltd. | Method, system and equipment for automatically detecting and tracking blade |
| CN111696075A (en) * | 2020-04-30 | 2020-09-22 | 航天图景（北京）科技有限公司 | Intelligent fan blade defect detection method based on double-spectrum image |
| WO2022148109A1 (en) * | 2021-01-05 | 2022-07-14 | 歌尔股份有限公司 | Product defect detection method and apparatus, device and computer-readable storage medium |
| CN112802011A (en) * | 2021-02-25 | 2021-05-14 | 上海电机学院 | Fan blade defect detection method based on VGG-BLS |
| CN113837994A (en) * | 2021-07-29 | 2021-12-24 | 尚特杰电力科技有限公司 | Photovoltaic panel defect diagnosis method based on edge detection convolutional neural network |
| CN113409314A (en) * | 2021-08-18 | 2021-09-17 | 南京市特种设备安全监督检验研究院 | Unmanned aerial vehicle visual detection and evaluation method and system for corrosion of high-altitude steel structure |
| EP4227856A1 (en) * | 2022-02-14 | 2023-08-16 | Samsung Display Co., Ltd. | System and method for defect detection |
| CN114663376A (en) * | 2022-03-15 | 2022-06-24 | 中国华能集团清洁能源技术研究院有限公司 | Fan blade defect detection method and system based on improved SSD model |
| CN115166032A (en) * | 2022-05-23 | 2022-10-11 | 东南大学 | Device and method for detecting cracks of fan blade |
| CN115937114A (en) * | 2022-11-25 | 2023-04-07 | 无锡海纳智能科技有限公司 | Fan picture preprocessing method and device |
| CN116206094A (en) * | 2023-04-28 | 2023-06-02 | 尚特杰电力科技有限公司 | Fan blade angle measuring method, device and system and electronic equipment |
| CN116342607A (en) * | 2023-05-30 | 2023-06-27 | 尚特杰电力科技有限公司 | Power transmission line defect identification method and device, electronic equipment and storage medium |
| CN116797594A (en) * | 2023-07-10 | 2023-09-22 | 华北电力大学（保定） | Fan blade defect detection method, system, electronic equipment and storage medium |
| CN116823872A (en) * | 2023-08-25 | 2023-09-29 | 尚特杰电力科技有限公司 | Fan inspection method and system based on target tracking and image segmentation |
| CN117058396A (en) * | 2023-10-11 | 2023-11-14 | 精效悬浮(苏州)科技有限公司 | Fan blade defect area rapid segmentation method and system based on artificial intelligence |
Non-Patent Citations (2)
| Title |
|---|
| 朱佳伟 等: "复合材料科学与工程", 《基于改进SSD的风机叶片缺陷检测》, no. 3, pages 38 - 44 * |
| 李建 等: "基于Transformer的目标检测算法综述", 《计算机工程与应用》, vol. 59, no. 10, pages 48 - 64 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117351062B (en) | 2024-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11361428B1 (en) | Technology for analyzing images depicting vehicles according to base image models | |
| US9886771B1 (en) | Heat map of vehicle damage | |
| US10319094B1 (en) | Technology for capturing, transmitting, and analyzing images of objects | |
| US10706321B1 (en) | Image processing system to align a target object in a target object image with an object model | |
| US11288789B1 (en) | Systems and methods for repairing a damaged vehicle using image processing | |
| US10636148B1 (en) | Image processing system to detect contours of an object in a target object image | |
| CN110148130B (en) | Method and device for detecting part defects | |
| US8331650B2 (en) | Methods, systems and apparatus for defect detection | |
| CN109376591B (en) | Ship target detection method for deep learning feature and visual feature combined training | |
| CN111428748A (en) | Infrared image insulator recognition and detection method based on HOG characteristics and SVM | |
| CN111381579A (en) | Cloud deck fault detection method and device, computer equipment and storage medium | |
| CN110634137A (en) | Bridge deformation monitoring method, device and equipment based on visual perception | |
| CN115359239A (en) | Wind power blade defect detection and positioning method and device, storage medium and electronic equipment | |
| CN113688817A (en) | Instrument identification method and system for automatic inspection | |
| CN115240089A (en) | Vehicle detection method of aerial remote sensing image | |
| CN114821274A (en) | Method and device for identifying state of split and combined indicator | |
| CN112565615B (en) | Method and device for determining trigger point of flying shooting | |
| CN112884795A (en) | Power transmission line inspection foreground and background segmentation method based on multi-feature significance fusion | |
| CN117351062B (en) | Fan blade defect diagnosis method, device and system and electronic equipment | |
| JP4270254B2 (en) | Image signal processing apparatus and image processing method | |
| CN117314986A (en) | Unmanned aerial vehicle cross-mode power distribution equipment inspection image registration method based on semantic segmentation | |
| JP3500425B2 (en) | Road surface condition judgment method in visible image type road surface condition grasping device | |
| Hoover et al. | Range image segmentation: The user's dilemma | |
| CN116046790A (en) | Defect detection method, device, system, electronic equipment and storage medium | |
| CN115619796A (en) | Method and device for obtaining photovoltaic module template and nonvolatile storage medium |
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 |