CN112727705A - Monitoring and flaw detection method for blades of wind generating set - Google Patents
Monitoring and flaw detection method for blades of wind generating set Download PDFInfo
- Publication number
- CN112727705A CN112727705A CN202011544911.2A CN202011544911A CN112727705A CN 112727705 A CN112727705 A CN 112727705A CN 202011544911 A CN202011544911 A CN 202011544911A CN 112727705 A CN112727705 A CN 112727705A
- Authority
- CN
- China
- Prior art keywords
- image
- blade
- point
- camera
- coordinate system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 25
- 238000001514 detection method Methods 0.000 title claims abstract description 15
- 230000007547 defect Effects 0.000 claims abstract description 50
- 238000000034 method Methods 0.000 claims abstract description 34
- 238000001914 filtration Methods 0.000 claims abstract description 14
- 239000004576 sand Substances 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims description 24
- 230000007797 corrosion Effects 0.000 claims description 16
- 238000005260 corrosion Methods 0.000 claims description 16
- 238000003384 imaging method Methods 0.000 claims description 13
- 230000010354 integration Effects 0.000 claims description 10
- 238000006073 displacement reaction Methods 0.000 claims description 9
- 238000013519 translation Methods 0.000 claims description 9
- 230000000007 visual effect Effects 0.000 claims description 8
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000003672 processing method Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 241000226585 Antennaria plantaginifolia Species 0.000 claims description 3
- 230000010339 dilation Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000000877 morphologic effect Effects 0.000 claims description 3
- 230000003287 optical effect Effects 0.000 claims description 3
- 230000003071 parasitic effect Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000004804 winding Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 4
- 238000012360 testing method Methods 0.000 abstract description 2
- 208000027418 Wounds and injury Diseases 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 239000003365 glass fiber Substances 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 208000025274 Lightning injury Diseases 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000000919 ceramic Substances 0.000 description 1
- 239000011248 coating agent Substances 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 150000003839 salts Chemical class 0.000 description 1
- 239000007921 spray Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
- F03D—WIND MOTORS
- F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
Abstract
The invention discloses a blade monitoring and flaw detection method for a wind generating set, which uses an unmanned aerial vehicle camera calibration method, a rapid self-adaptive weighted median filtering algorithm, an image enhancement algorithm, blade surface fault feature extraction and selection and blade surface fault classification monitoring and identification aiming at the characteristic that an unmanned aerial vehicle acquires images for a fan blade, and the unmanned aerial vehicle camera calibration method, the rapid self-adaptive weighted median filtering algorithm, the image enhancement algorithm, the blade surface fault feature extraction and selection and the blade surface fault classification monitoring and identification are embedded into a man-. The platform can realize automatic identification and monitoring of defects such as sand holes, cracks, peeling and the like through online and offline tests, the accuracy can reach more than 90%, and the platform has higher accuracy and better algorithm stability compared with the traditional monitoring means and other monitoring algorithms, and provides a new way for nondestructive monitoring of the wind driven generator blades.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a monitoring and flaw detection method for a blade of a wind generating set.
Background
The blade in the wind driven generator is a key component for capturing wind energy of the fan, and the fault clearing time of the blade accounts for the highest percentage in the whole fault clearing time of the fan. The wind field is mainly distributed in the suburbs, offshore areas, gobi and other geographical environments of cities and wind energy storage places with very complex climatic environments, so that the blades of the wind driven generator cannot be impacted by various disaster weathers such as hurricanes, salt sprays and the like; the surface of the blade is inevitably damaged by various media to form various defect damages. If these small defect lesions are not attended to and handled in a timely manner, they accumulate to form large faults, which eventually become serious accidents.
The statistical finding of the regular and routine monitoring results of the blade shows that the surface defects of the blade mainly comprise: spots and scratches appear in the early stage of fan installation, and sand holes, cracks, edge corrosion and cracking, surface coating falling off, lightning damage and blade icing in severe weather appear in the middle and later stages of fan installation; in general, spots, scratches, blisters and cracks are common defects of fan blades, and large-area spots and blisters are gathered to form pitted surfaces with greatly increased harmfulness. In the weather of thunderstorm, the electrical conductivity of the pitted surface is enhanced due to much electrostatic ash and high humidity, so that lightning stroke is easy to be induced; when the glass fiber is exposed to high temperature, strong wind and other weather, the glass fiber at the concave part of the pitted surface can be quickly weathered to form a brittle layer. The two conditions can cause serious damage to the blades, and even the condition of stopping and overhauling the fan occurs.
The surface monitoring and defect identification system of the fan blade based on the vibration sensor, the acoustic emission sensor, the strain sensor and the ultrasonic sensor is used, the piezoelectric ceramic sensor piece combined on the blade is used for exciting the vibration response of the blade, and the damage of the blade is monitored by various methods such as transmission function, resonance comparison, operation distortion shape and wave propagation.
Disclosure of Invention
The invention provides a blade monitoring and flaw detection method for a wind generating set, which is characterized in that an unmanned aerial vehicle camera calibration method, a rapid self-adaptive weighted median filtering processing image and a dynamic threshold segmentation blade image defect feature method are used for detecting defects such as cracks and sand holes in area processing and recognition, classifying and measuring the defects and outputting an analysis report of blade quality, so that the automatic detection function of the surface defects of the blade of the wind generating set is realized.
The technical scheme of the invention is as follows:
the blade monitoring and flaw detection method of the wind generating set comprises the steps of firstly, acquiring a surface image of a blade of the wind generating set by a camera carried on an unmanned aerial vehicle, then transmitting the acquired blade defect image to a ground computer in real time through a wireless network, and processing and defect identification on the acquired image by the ground computer by using a vision integration platform;
the specific steps of the visual integration platform for image processing and defect identification are as follows:
(1) calibrating the position and the form of the acquired image in a two-dimensional image coordinate by adopting an unmanned aerial vehicle camera calibration method; from a spatial point PWThe projection point P transformed onto the image plane is subjected to the following steps:
a. by changing the relation P by relative positionC＝RPW+ T to capture the object point P in the imageWConversion to point P in the camera coordinate systemCWhere R ═ (α, β, γ) is a rotation matrix, and T ═ T (T)x,ty,tz) Is a translation vector, alpha, beta, gamma being respectively a winding xw，yw，zwAngle of rotation of three coordinate axes, tx,ty,tzAre respectively wound around xw，yw，zwThe translation values of the three coordinate axes, wherein parameters in R and T are external parameters of the camera;
b. through a projection relation between a camera coordinate system and a pixel coordinate system of the image:
handle point PCConverting from the camera coordinate system to the pixel coordinate system of the image, where m, n are the pixel coordinate system of the image, xc，yc，zcIs a camera coordinate system, and f is the principal distance of the camera;
c. physical point PWA line lying in both the world coordinate system and a projected point P lying in the imaging plane will traverse the central optical position of the imaging device and perform a phase comparison using the radial distortion formula, i.e.:
wherein k is a distortion coefficient of the lens,
distortion coordinates in a pixel coordinate system of the image;
d. when the parameter k in the formula (2) is a negative number, the distortion is barrel distortion, and when the parameter k is a positive number, the distortion is pincushion distortion; this distortion is corrected by the following equation (3):
e. and finally, converting the distortion point into a physical model coordinate system of the image:
in formula (4): p, d is the physical model coordinate system of the image, cxAnd cyIs a vertical projection of the projection center on the imaging plane; sxAnd syThe length of the pixels in the image-sensing element in the transverse and radial directions between phases is then (f, k, c)x,cy,sx,sy) The method is characterized in that the method is internal parameters of a camera and determines the conversion relation of points in the camera;
(2) and preprocessing a blade surface fault image:
the uninterrupted image information has parasitic effect and interference noise, the image quality is reduced, and the unmanned aerial vehicle has jitter when carrying a camera to collect images, so that the blade images need to be denoised; based on noise factors of an unmanned aerial vehicle carrying camera during image acquisition and real-time requirements of rapidly processing images in the flight process, image denoising is carried out by adopting a rapid self-adaptive weighted median filtering method:
defining S (i, j) as an original image and an impulse noise contaminated image, where y (i, j) e Ω ═ 1, … M } × {1, … N }, M and N are width and height of the image, respectively, and a gray scale range of the image is [ S, j [ ]min,smax]Then the observed image is given by:
when r (i, j) takes the value sminOr smaxThen, impulse noise is formed; when r (i, j) is equal to [ s ]min,smax]Then random noise is formed;
definition sw(i, j) is a square window of size w × w centered at point (i, j), which can be expressed as:
sw i,j＝{(k,l)│|k-i|＜w,|i-j|＜w} (6)；
the specific method comprises the following steps:
let the maximum window be wmax×wmaxThe output image is u (i, j), the basic idea of the self-adaptive median filtering algorithm is to adjust the size of a window by judging whether the center point of the window is noise or not so as to overcome the damage of median filtering to details;
the specific algorithm is as follows:
a. initializing the window size, and enabling w to be 3;
b. calculating the window sw i,jMinimum value s of middle pixelmin,w i,jMaximum value smax,w i,jAnd the median value smed,w i,j；
c. If s ismin,w i,j＜smed,w i,j＜smax,w i,jSkipping to the step e; otherwise, increasing the window to w as w + 2;
d. if w < wmaxReturning to the step b; otherwise use the median smed,w i,jReplacing the current point y (i, j);
e. if s ismin,w i,j＜y(i,j)＜smax,w i,jIf the point is not noise, the output is kept unchanged; otherwise, the point is indicated as noise, and the median s is usedmed,w i,jTo replace the current point y (i, j);
(3) carrying out image enhancement and dynamic threshold segmentation on the denoised blade image:
the method adopts a multi-azimuth multi-structural-element processing method to enable the structural elements to cover all directions of the image, thereby achieving the purpose of better keeping the detailed information of the image; the purpose of dynamic threshold segmentation of an image is to divide the image into a plurality of statistically uniform mutually disjoint sub-regions or a set of divisions according to different characteristics of the image, wherein the set can be used as a target region and can be separated from an image background or other target regions; the multi-azimuth multi-structure processing method comprises the following specific steps:
a. and (3) corrosion operation:
for ZnThe above element sets A and B, B corrosion A, is written as A Θ B ═ z | (B)ZE is A }; with the structural element B at the origin over the whole ZnPlane movement; if the origin of B is translated to the point Z, B can be completely contained in A, and the set formed by all the points Z is the corrosion image of B to A;
b. and (3) expansion operation:
for ZnThe sets A and B of the upper elements, B performs the dilation operation on A and is recorded as
Mapping with B origin relative to itself
And based on the translation of the z-image, the expansion of B on A is understood to be
If at least one element in the displacement Z points is overlapped with at least one element in the displacement A, the set formed by the displacement Z points is the expansion image of the B pairs A;
c. opening and closing operation:
the open operation of set B on set A is represented as
Record as
The open operation of the set B on the set A is the corrosion of the set B on the set A, and then the result is expanded by the set B;
the closed operation of set B on set A, denoted A.B, is written as
The closed operation of the set B on the set A is the expansion of the set B on the set A, and then the set B corrodes the result;
opening and closing operations in a morphological image enhancement algorithm enable the image contour to become smooth; however, the opening operation mainly comprises the steps of disconnecting narrow connection, eliminating 'thin burrs' and removing false edges and false details; the closing operation can usually close narrow discontinuities and fill small holes;
(4) and extracting and selecting the surface fault characteristics of the enhanced blade image:
establishing a 12-feature-value feature pool containing image geometric features, textural features and gray-scale features, and selecting a classifier input with 6 feature values according to feature dimension disasters and the specific requirements of the classifier, so as to highlight edge information in a paddle image;
(5) carrying out blade surface fault classification detection and identification according to edge information in the protruding blade graph:
the method comprises the following steps that surface faults of the wind driven generator blade are mainly divided into cracks and corrosion surfaces, images are subjected to combination and expansion processing, a Selected Regions (Region Union) operator is used for containing pictures of Regions to be calculated of all the Regions and returning a set containing all the Regions, and then a Region of a non-square state unit is subjected to expansion processing by taking a circle as a template through a dimension _ circle (Region, Region division, Radius);
extracting a core line of crack characteristics on a blade, extracting a Skeleton of a main line of the crack by using a skelon Region division, skele (ton) operator, then performing interconnection processing on a joint Region of the crack by using a connection (skelon, Errors) operator, and then performing further processing on a heavy point target;
measuring the sizes and determining the types of grains of the cracks and the sand hole defects, firstly selecting a defect area by using a select _ shape (Regions: selected Regions: Fres, Operation, Min, Max:) operator according to a specific selection value, selecting areas as the characteristics, namely the number of pixels, wherein the sand hole defects are defined as 20-100 pixels, and the crack defects are defined as 300-9000 pixels; because the region is a non-square annular region, a skeleton operator skeeleton is operated to obtain a skeleton of the non-square annular region, then a gen _ controls _ skeeleton _ xld operator is used for generating an edge core of the skeleton, and finally a length _ xld (contents, length) operator is used for obtaining length-related data of the crack, so that the defect in the blade map can be marked.
And the ground computer sends the acquired blade defect image to a database of the memory for storage, the blade defect information processed and identified by the visual integration platform is also stored in the database of the memory, and then the blade defect image and the blade defect information are extracted from the database for defect playback and blade quality analysis report output is obtained.
In the step (4), the 12 characteristic values are area, perimeter, rectangle degree, Euler number, average gray level, gray level variance, gray level entropy, energy, entropy, correlation, contrast and contrast; after 6 eigenvalues are taken as inputs, the result output by the classifier is the prediction accuracy.
The unmanned aerial vehicle on carry on the camera for long-range cloud platform camera that zooms.
The invention has the advantages that:
aiming at the characteristics of an unmanned aerial vehicle for collecting images of fan blades, the invention uses an unmanned aerial vehicle camera calibration method, a rapid self-adaptive weighted median filtering algorithm, an image enhancement algorithm, blade surface fault feature extraction and selection, and blade surface fault classification monitoring and identification, and embeds the unmanned aerial vehicle camera calibration method, the rapid self-adaptive weighted median filtering algorithm, the image enhancement algorithm, the blade surface fault feature extraction and selection and the blade surface fault classification monitoring and identification into a human-computer interaction platform. The platform can realize automatic identification and monitoring of defects such as sand holes, cracks, peeling and the like through online and offline tests, the accuracy can reach more than 90%, and the platform has higher accuracy and better algorithm stability compared with the traditional monitoring means and other monitoring algorithms, and provides a new way for nondestructive monitoring of the wind driven generator blades.
Drawings
FIG. 1 is a process of image acquisition and processing of a wind turbine blade.
FIG. 2 is a flow chart of image processing and defect identification by the vision integration platform.
Fig. 3 is a camera imaging model built based on the pinhole imaging principle.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
According to the blade monitoring and flaw detection method of the wind generating set, firstly, a remote zooming pan-tilt camera carried on an unmanned aerial vehicle is adopted to acquire images of the surface of a blade of the wind generating set, then the acquired images of the blade defect are transmitted to a ground computer in real time through a wireless network, the ground computer uses a visual integration platform to process and identify the acquired images of the blade defect, the ground computer sends the acquired images of the blade defect to a database of a memory to be stored, the blade defect information processed and identified by the visual integration platform is also stored in the database of the memory, and then the images of the blade defect and the blade defect information are extracted from the database to be played back in defect, and a blade quality analysis report is prepared and output;
referring to fig. 2, the specific steps of the visual integration platform for image processing and defect identification are as follows:
(1) calibrating the position and the form of the acquired image in a two-dimensional image coordinate by adopting an unmanned aerial vehicle camera calibration method; a camera imaging model established based on the pinhole imaging principle is shown in fig. 3; from a spatial point PWThe projection point P transformed onto the image plane is subjected to the following steps:
a. by changing the relation P by relative positionC＝RPW+ T to capture the object point P in the imageWConversion to point P in the camera coordinate systemCWhere R ═ (α, β, γ) is a rotation matrix, and T ═ T (T)x,ty,tz) Is a translation vector, alpha, beta, gamma being respectively a winding xw，yw，zwAngle of rotation of three coordinate axes, tx,ty,tzAre respectively wound around xw，yw，zwThe translation values of the three coordinate axes, wherein parameters in R and T are external parameters of the camera;
b. through a projection relation between a camera coordinate system and a pixel coordinate system of the image:
handle point PCConverting from the camera coordinate system to the pixel coordinate system of the image, where m, n are the pixel coordinate system of the image, xc，yc，zcIs a camera coordinate system, and f is the principal distance of the camera;
c. physical point PWA line lying in both the world coordinate system and a projected point P lying in the imaging plane will traverse the central optical position of the imaging device and perform a phase comparison using the radial distortion formula, i.e.:
wherein k is a distortion coefficient of the lens,
as an imageDistortion coordinates in the pixel coordinate system;
d. when the parameter k in the formula (2) is a negative number, the distortion is barrel distortion, and when the parameter k is a positive number, the distortion is pincushion distortion; this distortion is corrected by the following equation (3):
e. and finally, converting the distortion point into a physical model coordinate system of the image:
in formula (4): p, d is the physical model coordinate system of the image, cxAnd cyIs a vertical projection of the projection center on the imaging plane; sxAnd syThe length of the pixels in the image-sensing element in the transverse and radial directions between phases is then (f, k, c)x,cy,sx,sy) The method is characterized in that the method is internal parameters of a camera and determines the conversion relation of points in the camera;
(2) and preprocessing a blade surface fault image:
the uninterrupted image information has parasitic effect and interference noise, the image quality is reduced, and the unmanned aerial vehicle has jitter when carrying a camera to collect images, so that the blade images need to be denoised; based on noise factors of an unmanned aerial vehicle carrying camera during image acquisition and real-time requirements of rapidly processing images in the flight process, image denoising is carried out by adopting a rapid self-adaptive weighted median filtering method:
defining S (i, j) as an original image and an impulse noise contaminated image, where y (i, j) e Ω ═ 1, … M } × {1, … N }, M and N are width and height of the image, respectively, and a gray scale range of the image is [ S, j [ ]min,smax]Then the observed image is given by:
when r (i, j) takes the value sminOr smaxThen, impulse noise is formed; when r (i, j) is equal to [ s ]min,smax]Then random noise is formed;
definition sw(i, j) is a square window of size w × w centered at point (i, j), which can be expressed as:
sw i,j＝{(k,l)│|k-i|＜w,|i-j|＜w} (6)；
the specific method comprises the following steps:
let the maximum window be wmax×wmaxThe output image is u (i, j), the basic idea of the self-adaptive median filtering algorithm is to adjust the size of a window by judging whether the center point of the window is noise or not so as to overcome the damage of median filtering to details;
the specific algorithm is as follows:
a. initializing the window size, and enabling w to be 3;
b. calculating the window sw i,jMinimum value s of middle pixelmin,w i,jMaximum value smax,w i,jAnd the median value smed,w i,j；
c. If s ismin,w i,j＜smed,w i,j＜smax,w i,jSkipping to the step e; otherwise, increasing the window to w as w + 2;
d. if w < wmaxReturning to the step b; otherwise use the median smed,w i,jReplacing the current point y (i, j);
e. if s ismin,w i,j＜y(i,j)＜smax,w i,jIf the point is not noise, the output is kept unchanged; otherwise, the point is indicated as noise, and the median s is usedmed,w i,jTo replace the current point y (i, j);
(3) carrying out image enhancement and dynamic threshold segmentation on the denoised blade image:
the method adopts a multi-azimuth multi-structural-element processing method to enable the structural elements to cover all directions of the image, thereby achieving the purpose of better keeping the detailed information of the image; the purpose of dynamic threshold segmentation of an image is to divide the image into a plurality of statistically uniform mutually disjoint sub-regions or a set of divisions according to different characteristics of the image, wherein the set can be used as a target region and can be separated from an image background or other target regions; the multi-azimuth multi-structure processing method comprises the following specific steps:
a. and (3) corrosion operation:
for ZnThe above element sets A and B, B corrosion A, is written as A Θ B ═ z | (B)ZE is A }; with the structural element B at the origin over the whole ZnPlane movement; if the origin of B is translated to the point Z, B can be completely contained in A, and the set formed by all the points Z is the corrosion image of B to A;
b. and (3) expansion operation:
for ZnThe sets A and B of the upper elements, B performs the dilation operation on A and is recorded as
Mapping with B origin relative to itself
And based on the translation of the z-image, the expansion of B on A is understood to be
If at least one element in the displacement Z points is overlapped with at least one element in the displacement A, the set formed by the displacement Z points is the expansion image of the B pairs A;
c. opening and closing operation:
the open operation of set B on set A is represented as
Record as
The open operation of the set B on the set A is the corrosion of the set B on the set A, and then the result is expanded by the set B;
closed operation of set B on set AA and B are expressed as
The closed operation of the set B on the set A is the expansion of the set B on the set A, and then the set B corrodes the result;
opening and closing operations in a morphological image enhancement algorithm enable the image contour to become smooth; however, the opening operation mainly comprises the steps of disconnecting narrow connection, eliminating 'thin burrs' and removing false edges and false details; the closing operation can usually close narrow discontinuities and fill small holes;
(4) and extracting and selecting the surface fault characteristics of the enhanced blade image:
establishing a 12-feature-value feature pool containing image geometric features, textural features and gray-scale features, and selecting a classifier input with 6 feature values according to feature dimension disasters and the specific requirements of the classifier, so as to highlight edge information in a paddle image; the 12 characteristic values are area, perimeter, rectangle degree, Euler number, average gray scale, gray scale variance, gray scale entropy, energy, entropy, correlation, contrast and contrast; after the characteristic value is taken as input, the result output by the classifier is the prediction accuracy;
(5) carrying out blade surface fault classification detection and identification according to edge information in the protruding blade graph:
the method comprises the following steps that surface faults of the wind driven generator blade are mainly divided into cracks and corrosion surfaces, images are subjected to combination and expansion processing, a Selected Regions (Region Union) operator is used for containing pictures of Regions to be calculated of all the Regions and returning a set containing all the Regions, and then a Region of a non-square state unit is subjected to expansion processing by taking a circle as a template through a dimension _ circle (Region, Region division, Radius);
extracting a core line of crack characteristics on a blade, extracting a Skeleton of a main line of the crack by using a skelon Region division, skele (ton) operator, then performing interconnection processing on a joint Region of the crack by using a connection (skelon, Errors) operator, and then performing further processing on a heavy point target;
measuring the sizes and determining the types of grains of the cracks and the sand hole defects, firstly selecting a defect area by using a select _ shape (Regions: selected Regions: Fres, Operation, Min, Max:) operator according to a specific selection value, selecting areas as the characteristics, namely the number of pixels, wherein the sand hole defects are defined as 20-100 pixels, and the crack defects are defined as 300-9000 pixels; because the region is a non-square annular region, a skeleton operator skeeleton is operated to obtain a skeleton of the non-square annular region, then a gen _ controls _ skeeleton _ xld operator is used for generating an edge core of the skeleton, and finally a length _ xld (contents, length) operator is used for obtaining length-related data of the crack, so that the defect in the blade map can be marked.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (4)
1. The monitoring flaw detection method for the blades of the wind generating set is characterized by comprising the following steps of: firstly, acquiring an image of the surface of a blade of a wind turbine generator by using a camera carried on an unmanned aerial vehicle, and then transmitting the acquired image of the blade defect to a ground computer in real time through a wireless network, wherein the ground computer processes and identifies the acquired image by using a visual integration platform;
the specific steps of the visual integration platform for image processing and defect identification are as follows:
(1) calibrating the position and the form of the acquired image in a two-dimensional image coordinate by adopting an unmanned aerial vehicle camera calibration method; from a spatial point PWThe projection point P transformed onto the image plane is subjected to the following steps:
a. by changing the relation P by relative positionC＝RPW+ T to capture the object point P in the imageWConversion to point P in the camera coordinate systemCWhere R ═ (α, β, γ) is a rotation matrix, and T ═ T (T)x,ty,tz) Is a translation vector, αBeta, gamma are each a winding xw，yw，zwAngle of rotation of three coordinate axes, tx,ty,tzAre respectively wound around xw，yw，zwThe translation values of the three coordinate axes, wherein parameters in R and T are external parameters of the camera;
b. through a projection relation between a camera coordinate system and a pixel coordinate system of the image:
handle point PCConverting from the camera coordinate system to the pixel coordinate system of the image, where m, n are the pixel coordinate system of the image, xc，yc，zcIs a camera coordinate system, and f is the principal distance of the camera;
c. physical point PWA line lying in both the world coordinate system and a projected point P lying in the imaging plane will traverse the central optical position of the imaging device and perform a phase comparison using the radial distortion formula, i.e.:
wherein k is a distortion coefficient of the lens,
distortion coordinates in a pixel coordinate system of the image;
d. when the parameter k in the formula (2) is a negative number, the distortion is barrel distortion, and when the parameter k is a positive number, the distortion is pincushion distortion; this distortion is corrected by the following equation (3):
e. and finally, converting the distortion point into a physical model coordinate system of the image:
in formula (4): p, d is the physical model coordinate system of the image, cxAnd cyIs a vertical projection of the projection center on the imaging plane; sxAnd syThe length of the pixels in the image-sensing element in the transverse and radial directions between phases is then (f, k, c)x,cy,sx,sy) The method is characterized in that the method is internal parameters of a camera and determines the conversion relation of points in the camera;
(2) and preprocessing a blade surface fault image:
the uninterrupted image information has parasitic effect and interference noise, the image quality is reduced, and the unmanned aerial vehicle has jitter when carrying a camera to collect images, so that the blade images need to be denoised; based on noise factors of an unmanned aerial vehicle carrying camera during image acquisition and real-time requirements of rapidly processing images in the flight process, image denoising is carried out by adopting a rapid self-adaptive weighted median filtering method:
defining S (i, j) as an original image and an impulse noise contaminated image, where y (i, j) e Ω ═ 1, … M } × {1, … N }, M and N are width and height of the image, respectively, and a gray scale range of the image is [ S, j [ ]min,smax]Then the observed image is given by:
when r (i, j) takes the value sminOr smaxThen, impulse noise is formed; when r (i, j) is equal to [ s ]min,smax]Then random noise is formed;
definition sw(i, j) is a square window of size w × w centered at point (i, j), which can be expressed as:
sw i,j＝{(k,l)│|k-i|＜w,|i-j|＜w} (6)；
the specific method comprises the following steps:
let the maximum window be wmax×wmaxThe output image is u (i, j), the basic idea of the self-adaptive median filtering algorithm is to adjust the size of a window by judging whether the center point of the window is noise or not so as to overcome the damage of median filtering to details;
the specific algorithm is as follows:
a. initializing the window size, and enabling w to be 3;
b. calculating the window sw i,jMinimum value s of middle pixelmin,w i,jMaximum value smax,w i,jAnd the median value smed,w i,j；
c. If s ismin,w i,j＜smed,w i,j＜smax,w i,jSkipping to the step e; otherwise, increasing the window to w as w + 2;
d. if w < wmaxReturning to the step b; otherwise use the median smed,w i,jReplacing the current point y (i, j);
e. if s ismin,w i,j＜y(i,j)＜smax,w i,jIf the point is not noise, the output is kept unchanged; otherwise, the point is indicated as noise, and the median s is usedmed,w i,jTo replace the current point y (i, j);
(3) carrying out image enhancement and dynamic threshold segmentation on the denoised blade image:
the method adopts a multi-azimuth multi-structural-element processing method to enable the structural elements to cover all directions of the image, thereby achieving the purpose of better keeping the detailed information of the image; the purpose of dynamic threshold segmentation of an image is to divide the image into a plurality of statistically uniform mutually disjoint sub-regions or a set of divisions according to different characteristics of the image, wherein the set can be used as a target region and can be separated from an image background or other target regions; the multi-azimuth multi-structure processing method comprises the following specific steps:
a. and (3) corrosion operation:
for ZnThe above element sets A and B, B corrosion A, is written as A Θ B ═ z | (B)ZE is A }; with the structural element B at the origin over the whole ZnPlane movement; if the origin of B is translated to the point Z, B can be completely contained in A, and the set formed by all the points Z is the corrosion image of B to A;
b. and (3) expansion operation:
for ZnThe sets A and B of the upper elements, B performs the dilation operation on A and is recorded as
Mapping with B origin relative to itself
And based on the translation of the z-image, the expansion of B on A is understood to be
If at least one element in the displacement Z points is overlapped with at least one element in the displacement A, the set formed by the displacement Z points is the expansion image of the B pairs A;
c. opening and closing operation:
the open operation of set B on set A is represented as
Record as
The open operation of the set B on the set A is the corrosion of the set B on the set A, and then the result is expanded by the set B;
the closed operation of set B on set A, denoted A.B, is written as
The closed operation of the set B on the set A is the expansion of the set B on the set A, and then the set B corrodes the result;
opening and closing operations in a morphological image enhancement algorithm enable the image contour to become smooth; however, the opening operation mainly comprises the steps of disconnecting narrow connection, eliminating 'thin burrs' and removing false edges and false details; the closing operation can usually close narrow discontinuities and fill small holes;
(4) and extracting and selecting the surface fault characteristics of the enhanced blade image:
establishing a 12-feature-value feature pool containing image geometric features, textural features and gray-scale features, and selecting a classifier input with 6 feature values according to feature dimension disasters and the specific requirements of the classifier, so as to highlight edge information in a paddle image;
(5) carrying out blade surface fault classification detection and identification according to edge information in the protruding blade graph:
the method comprises the following steps that surface faults of the wind driven generator blade are mainly divided into cracks and corrosion surfaces, images are subjected to combination and expansion processing, a Selected Regions (Region Union) operator is used for containing pictures of Regions to be calculated of all the Regions and returning a set containing all the Regions, and then a Region of a non-square state unit is subjected to expansion processing by taking a circle as a template through a dimension _ circle (Region, Region division, Radius);
extracting a core line of crack characteristics on a blade, extracting a Skeleton of a main line of the crack by using a skelon Region division, skele (ton) operator, then performing interconnection processing on a joint Region of the crack by using a connection (skelon, Errors) operator, and then performing further processing on a heavy point target;
measuring the sizes and determining the types of grains of the cracks and the sand hole defects, firstly selecting a defect area by using a select _ shape (Regions: selected Regions: Fres, Operation, Min, Max:) operator according to a specific selection value, selecting areas as the characteristics, namely the number of pixels, wherein the sand hole defects are defined as 20-100 pixels, and the crack defects are defined as 300-9000 pixels; because the region is a non-square annular region, a skeleton operator skeeleton is operated to obtain a skeleton of the non-square annular region, then a gen _ controls _ skeeleton _ xld operator is used for generating an edge core of the skeleton, and finally a length _ xld (contents, length) operator is used for obtaining length-related data of the crack, so that the defect in the blade map can be marked.
2. The wind generating set blade monitoring flaw detection method according to claim 1, characterized in that: and the ground computer sends the acquired blade defect image to a database of the memory for storage, the blade defect information processed and identified by the visual integration platform is also stored in the database of the memory, and then the blade defect image and the blade defect information are extracted from the database for defect playback and blade quality analysis report output is obtained.
3. The wind generating set blade monitoring flaw detection method according to claim 1, characterized in that: in the step (4), the 12 characteristic values are area, perimeter, rectangle degree, Euler number, average gray level, gray level variance, gray level entropy, energy, entropy, correlation, contrast and contrast; after 6 eigenvalues are taken as inputs, the result output by the classifier is the prediction accuracy.
4. The wind generating set blade monitoring flaw detection method according to claim 1, characterized in that: the unmanned aerial vehicle on carry on the camera for long-range cloud platform camera that zooms.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011544911.2A CN112727705A (en) | 2020-12-23 | 2020-12-23 | Monitoring and flaw detection method for blades of wind generating set |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011544911.2A CN112727705A (en) | 2020-12-23 | 2020-12-23 | Monitoring and flaw detection method for blades of wind generating set |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112727705A true CN112727705A (en) | 2021-04-30 |
Family
ID=75605233
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011544911.2A Pending CN112727705A (en) | 2020-12-23 | 2020-12-23 | Monitoring and flaw detection method for blades of wind generating set |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112727705A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113284134A (en) * | 2021-06-17 | 2021-08-20 | 张清坡 | Unmanned aerial vehicle flight platform for geological survey |
| CN114719749A (en) * | 2022-04-06 | 2022-07-08 | 重庆大学 | Metal surface crack detection and real size measurement method and system based on machine vision |
| CN115875008A (en) * | 2023-01-06 | 2023-03-31 | 四川省川建勘察设计院有限公司 | Intelligent drilling data acquisition method and system for geological drilling machine and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102011017564A1 (en) * | 2011-04-26 | 2012-10-31 | Aerospy Sense & Avoid Technology Gmbh | Method and system for inspecting a surface for material defects |
| EP3543522A1 (en) * | 2018-03-22 | 2019-09-25 | Siemens Gamesa Renewable Energy A/S | Rotor blade monitoring system |
| CN111336073A (en) * | 2020-03-04 | 2020-06-26 | 南京航空航天大学 | Wind driven generator tower clearance visual monitoring device and method |
| CN111400961A (en) * | 2020-02-17 | 2020-07-10 | 通标标准技术服务有限公司 | Wind generating set blade fault judgment method and device |
| CN111852792A (en) * | 2020-09-10 | 2020-10-30 | 东华理工大学 | Fan blade defect self-diagnosis positioning method based on machine vision |
-
2020
- 2020-12-23 CN CN202011544911.2A patent/CN112727705A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102011017564A1 (en) * | 2011-04-26 | 2012-10-31 | Aerospy Sense & Avoid Technology Gmbh | Method and system for inspecting a surface for material defects |
| EP3543522A1 (en) * | 2018-03-22 | 2019-09-25 | Siemens Gamesa Renewable Energy A/S | Rotor blade monitoring system |
| CN111400961A (en) * | 2020-02-17 | 2020-07-10 | 通标标准技术服务有限公司 | Wind generating set blade fault judgment method and device |
| CN111336073A (en) * | 2020-03-04 | 2020-06-26 | 南京航空航天大学 | Wind driven generator tower clearance visual monitoring device and method |
| CN111852792A (en) * | 2020-09-10 | 2020-10-30 | 东华理工大学 | Fan blade defect self-diagnosis positioning method based on machine vision |
Non-Patent Citations (4)
| Title |
|---|
| 仇梓峰等: "基于无人机图像的风力发电机叶片缺陷识别", 《发电技术》 * |
| 王艳春等: "基于数学形态滤波算子的黄顶菊种子图像边缘检测", 《农机化研究》 * |
| 许国根等: "《模式识别与智能计算的MATLAB实现》", 31 August 2017, 北京航空航天大学出版社 * |
| 钱伟新等: "闪光照相CCD图像的自适应中值滤波方法", 《光学与光电技术》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113284134A (en) * | 2021-06-17 | 2021-08-20 | 张清坡 | Unmanned aerial vehicle flight platform for geological survey |
| CN113284134B (en) * | 2021-06-17 | 2023-09-26 | 张清坡 | Unmanned aerial vehicle flight platform for geological survey |
| CN114719749A (en) * | 2022-04-06 | 2022-07-08 | 重庆大学 | Metal surface crack detection and real size measurement method and system based on machine vision |
| CN114719749B (en) * | 2022-04-06 | 2023-07-14 | 重庆大学 | Metal surface crack detection and real size measurement method and system based on machine vision |
| CN115875008A (en) * | 2023-01-06 | 2023-03-31 | 四川省川建勘察设计院有限公司 | Intelligent drilling data acquisition method and system for geological drilling machine and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112727705A (en) | Monitoring and flaw detection method for blades of wind generating set | |
| WO2018028103A1 (en) | Unmanned aerial vehicle power line inspection method based on characteristics of human vision | |
| CN105976352B (en) | One kind being based on the morphologic face of weld defect characteristic extracting method of gray level image | |
| CN111339882B (en) | Power transmission line hidden danger detection method based on example segmentation | |
| CN113125458B (en) | Method and system for checking and evaluating coating state of steel structure | |
| CN113469177A (en) | Drainage pipeline defect detection method and system based on deep learning | |
| CN110503637B (en) | Road crack automatic detection method based on convolutional neural network | |
| Chen et al. | Automated bridge deck evaluation through UAV derived point cloud | |
| CN112750104B (en) | Method and device for automatically matching optimal camera by monitoring ship through multiple cameras | |
| CN112101138A (en) | Bridge inhaul cable surface defect real-time identification system and method based on deep learning | |
| US20220128358A1 (en) | Smart Sensor Based System and Method for Automatic Measurement of Water Level and Water Flow Velocity and Prediction | |
| CN114973032A (en) | Photovoltaic panel hot spot detection method and device based on deep convolutional neural network | |
| CN111683221B (en) | Real-time video monitoring method and system for natural resources embedded with vector red line data | |
| CN113822844A (en) | Unmanned aerial vehicle inspection defect detection method and device for blades of wind turbine generator system and storage medium | |
| CN113313107A (en) | Intelligent detection and identification method for multiple types of diseases on cable surface of cable-stayed bridge | |
| CN114973028B (en) | Aerial video image real-time change detection method and system | |
| Lee et al. | A new image-quality evaluating and enhancing methodology for bridge inspection using an unmanned aerial vehicle | |
| CN115497015A (en) | River floating pollutant identification method based on convolutional neural network | |
| CN114719749B (en) | Metal surface crack detection and real size measurement method and system based on machine vision | |
| CN115639248A (en) | System and method for detecting quality of building outer wall | |
| Siegel et al. | Remote enhanced visual inspection of aircraft by a mobile robot | |
| CN111524121A (en) | Road and bridge fault automatic detection method based on machine vision technology | |
| CN111597904A (en) | Identification method for inclination of tunnel cable support | |
| CN114359149A (en) | Dam bank dangerous case video detection method and system based on real-time image edge enhancement | |
| CN113160209A (en) | Target marking method and target identification method for building facade damage detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210430 |