CN115965623A - Surface flaw detection method and system in transformer production - Google Patents
Surface flaw detection method and system in transformer production Download PDFInfo
- Publication number
- CN115965623A CN115965623A CN202310251600.4A CN202310251600A CN115965623A CN 115965623 A CN115965623 A CN 115965623A CN 202310251600 A CN202310251600 A CN 202310251600A CN 115965623 A CN115965623 A CN 115965623A
- Authority
- CN
- China
- Prior art keywords
- value
- crack
- flaw
- pixel point
- detected
- 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
Images
Abstract
The invention relates to the technical field of image data processing, in particular to a method and a system for detecting surface flaws in transformer production, wherein the method is used for acquiring surface images of all parts of a transformer bushing to obtain a flaw detection area in the surface images; combining the corresponding values and gray values of each pixel point in the flaw to-be-detected area in RGB three channels and the corresponding edge distance to obtain the azimuth gloss characteristic value of the flaw to-be-detected area; combining the corresponding goodness of fit, the number of angular points, the crack width and the first number of clusters of the to-be-detected-flaw area to obtain the hot crack tortuosity of the to-be-detected-flaw area; and combining the azimuth gloss characteristic value and the hot crack tortuosity of the to-be-detected-flaw area to obtain hot crack significance, and detecting the surface flaws of the corresponding parts according to the hot crack significance of each to-be-detected-flaw area. The invention improves the accuracy and convenience of detecting the surface flaws of each part of the transformer bushing and reduces the loss of manpower and resources.
Description
Technical Field
The invention relates to the technical field of image data processing, in particular to a method and a system for detecting surface flaws in transformer production.
Background
The transformer bushing is a current-carrying element of the transformer, is a main insulation device of the outgoing lines of the transformer winding, and has the functions of insulating the outgoing lines and the transformer shell and fixing the outgoing lines. In order to ensure the normal operation of the transformer and the durable use of the transformer bushing and prevent accidents such as ground short circuit and insulation fault of power equipment, the full-sealed vacuum inside the transformer bushing needs to be ensured, so that the air tightness of the transformer bushing needs to be detected in time.
At present, the airtightness of the transformer bushing is mainly detected by infrared detection, sound discrimination and an electrical experiment method, wherein the infrared detection is mainly sensitive to temperature change, the sound discrimination is easily influenced by the environment, so that the accuracy is insufficient, and the electrical experiment method has accurate results but has a complex process and consumes a large amount of manpower and material resources. These methods are mainly used to check the surface quality and the airtightness of the assembled transformer and are not applicable during the assembly and welding processes. If cracks appear on the surface of each part of the transformer bushing after production and before bushing assembly, the quality of the transformer bushing assembled by the parts is poor, the service life is shortened, and the sealing performance is poor; if the inspection is performed after the assembly is completed, a lot of time and raw materials are wasted, and therefore, a crack defect inspection method which can cause a reduction in service life during the production and assembly process of the transformer bushing is required.
Disclosure of Invention
In order to solve the above technical problems, the present invention aims to provide a method and a system for detecting surface flaws in transformer production, and the adopted technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a method for detecting surface defects in transformer production, where the method includes:
collecting surface images of all parts of a transformer bushing, wherein the surface images are RGB images, and obtaining surface gray level images corresponding to the RGB images; performing edge detection on the surface gray level image to obtain an area corresponding to a closed edge line as a flaw to-be-detected area;
acquiring the edge distance from any pixel point in the flaw to-be-detected area to the corresponding image edge of the surface image, and acquiring the azimuth gloss characteristic value of the flaw to-be-detected area by combining the corresponding value and gray value of each pixel point in the flaw to-be-detected area in RGB three channels and the corresponding edge distance;
performing linear fitting on all pixel points in the flaw to-be-detected area to obtain goodness of fit; carrying out corner detection on the surface gray level image, and counting the number of corners in a flaw to-be-detected area; thinning the flaw to-be-detected area to obtain a corresponding skeleton line, making a straight line for each pixel point on the skeleton line, and acquiring the crack width of the corresponding pixel point according to the intersection point of the straight line and the flaw to-be-detected area; clustering the pixel points based on the crack width to obtain a first number of clusters, and combining the goodness of fit, the number of angular points, the crack width and the first number of clusters to obtain the hot crack tortuosity of the flaw detection area;
combining the azimuth gloss characteristic value and the hot crack tortuosity of the flaw to-be-detected area to obtain the hot crack significance; and detecting the surface defects of the corresponding parts according to the hot crack significance of each defect region to be detected.
Further, the method for obtaining the edge distance from any one pixel point in the flaw to-be-detected area to the image edge corresponding to the surface image includes:
and taking any pixel point in the region to be detected of the flaw as a target pixel point, respectively calculating the difference value between the corresponding Manhattan distance between the target pixel point and any pixel point on the image edge corresponding to the surface image and the Chebyshev distance based on the coordinates of the target pixel point, and taking the minimum difference value as the edge distance from the target pixel point to the image edge corresponding to the surface image.
Further, the method for obtaining the azimuth gloss characteristic value of the flaw to-be-detected area by combining the corresponding values, the gray values and the corresponding edge distances of each pixel point in the flaw to-be-detected area in the RGB three channels comprises the following steps:
arranging the edge distances of all pixel points in the flaw to-be-detected area from small to large to obtain a sequence, and acquiring pixel points corresponding to the first M edge distances in the sequence as first pixel points; m is a positive integer;
calculating an average edge distance according to the edge distance of the first pixel point; the values corresponding to the RGB three channels are an R value, a B value and a G value, for each first pixel point, a first difference value between the maximum value and the R value, a second difference value between the maximum value and the B value and a third difference value between the maximum value and the G value are respectively calculated, the addition result of the first difference value, the second difference value and the third difference value is calculated, the addition result is used as a numerator, the gray value corresponding to the first pixel point is used as a denominator, and the corresponding ratio is obtained, and the ratios of all the first pixel points are added to obtain the ratio sum;
and taking the product of the reciprocal of the average edge distance and the sum of the ratios as the azimuth gloss characteristic value of the flaw detected area.
Further, the method for making a straight line for each pixel point on the skeleton line and obtaining the crack width of the corresponding pixel point according to the intersection point of the straight line and the flaw to-be-detected area includes:
numbering each pixel point on the skeleton line in sequence, wherein for the ith pixel point, i =1, 2 and 3 \ 8230n, N-1, N is the number of the pixel points on the skeleton line, and N is a positive integer; and connecting the ith pixel point and the (i + 1) th pixel point to obtain a straight line as an ith straight line, making a target straight line perpendicular to the ith straight line for the ith pixel point to obtain two intersection points of the target straight line in the flaw to-be-detected area, and taking the Euclidean distance between the two intersection points as the crack width of the ith pixel point.
Further, the method for clustering pixel points based on crack widths to obtain a first number of clusters includes:
and constructing a planar rectangular coordinate system by taking the serial number as an abscissa and the crack width of the pixel point corresponding to the serial number as an ordinate, and clustering all the pixel points by using a DBSCAN algorithm according to each coordinate point in the planar rectangular coordinate system to obtain clusters of a first quantity.
Further, the method for obtaining the hot crack tortuosity of the flaw detected area by combining the goodness of fit, the number of angular points, the crack width and the first number of clusters includes:
respectively obtaining a crack width mean value, a maximum crack width and a minimum crack width based on all crack widths; calculating a first difference absolute value of the maximum crack width and the minimum crack width; obtaining the absolute value of the difference value between the width of each crack and the mean value of the width of the crack, and obtaining an average absolute value of the difference value according to the absolute value of the difference value corresponding to the width of each crack;
and taking the product of the goodness of fit, the number of angular points, the first difference absolute value, the average difference absolute value and the first number of clusters as the hot crack tortuosity of the flaw detection area.
Further, the method for obtaining the hot crack significance by combining the azimuth gloss characteristic value and the hot crack tortuosity of the detected flaw area comprises the following steps:
and taking the product of the azimuth gloss characteristic value and the hot crack tortuosity of the flaw detected area as the hot crack significance.
Further, the method for detecting the surface defects of the corresponding component according to the hot cracking significance of each defect region to be detected comprises the following steps:
normalizing the hot crack significance of each defect to-be-detected area to obtain a corresponding hot crack significance normalized value, setting a normalized threshold value, and when the hot crack significance normalized value is greater than or equal to the normalized threshold value, determining that the corresponding defect to-be-detected area is hot cracks;
and setting a number threshold, counting the number of the detected areas with the defects confirmed as hot cracks, and confirming that the hot cracks exist on the surface of the corresponding part when the number is greater than or equal to the number threshold.
In a second aspect, another embodiment of the present invention provides a system for detecting surface defects in transformer production, the system including: a memory, a processor and a computer program stored in said memory and executable on said processor, said computer program, when executed by said processor, performing the steps of any of the methods described above.
The invention has the following beneficial effects: the method comprises the steps of collecting surface images of all parts of the transformer bushing, and taking an area corresponding to a closed edge line as a flaw to-be-detected area through the edge line in the surface images based on the closed edge characteristic of an area corresponding to a hot crack; because the flaw to-be-detected area is possibly a position where cracks occur to affect air tightness, in order to determine whether the flaw to-be-detected area is hot cracks or not, the position close to the edge of a casting can be known to easily occur based on the reason of the occurrence of the hot cracks, and then the azimuth gloss characteristic value of the flaw to-be-detected area is obtained by combining the edge distance between each pixel point in the flaw to-be-detected area and the image edge corresponding to the surface image and the values and gray values of each pixel point corresponding to RGB three channels; in order to further accurately confirm whether the to-be-detected flaw area is a thermal flaw or not, based on the characteristics of strips, zigzag twisting and uneven inner shape and thickness generated by the thermal flaw, linear fitting and refining are carried out on the to-be-detected flaw area to obtain fitting goodness and skeleton lines, the first number of crack widths and cluster clusters is further obtained according to the skeleton lines, the number of corner points of the to-be-detected flaw area is detected at the same time for representing the zigzag phenomenon, the hot flaw tortuosity of the to-be-detected flaw area is obtained by combining the fitting goodness, the number of corner points, the crack width and the first number of cluster, and the hot flaw tortuosity measures the tortuosity and the width nonuniformity of the thermal flaw corresponding to the to-be-detected flaw area, and is more zigzag, more uneven in width and more likely to be a thermal flaw; the azimuth gloss characteristic value and the hot crack tortuosity of the flaw to-be-detected area are used as two characteristic indexes for judging whether the flaw is hot crack or not, so that the obtained hot crack significance is more rigorous and effective, the accuracy and convenience for detecting the surface flaws of the corresponding parts according to the hot crack significance are improved, and meanwhile, the loss of manpower and resources is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flowchart illustrating steps of a method for detecting surface defects in transformer manufacturing according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating skeleton lines of a defective test area according to an embodiment of the present disclosure.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the method and system for detecting surface defects in transformer production according to the present invention with reference to the accompanying drawings and preferred embodiments, and the detailed implementation, structure, features and effects thereof are described below. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The specific scenes aimed by the invention are as follows: in the production process of each part of the transformer bushing, the forged parts corresponding to the manufactured parts are subjected to image acquisition, the images are processed and analyzed, and the crack parts are identified, so that the air tightness of the whole bushing is ensured, and the inside of the bushing can be kept in vacuum in the working process.
The following describes a specific scheme of a method and a system for detecting surface flaws in transformer production in detail with reference to the accompanying drawings.
Referring to fig. 1, a flowchart illustrating steps of a method for detecting surface defects in transformer production according to an embodiment of the present invention is shown, where the method includes:
s001, collecting surface images of all parts of the transformer bushing, wherein the surface images are RGB images, and obtaining surface gray level images corresponding to the RGB images; and performing edge detection on the surface gray level image to obtain an area corresponding to the closed edge line as a flaw to-be-detected area.
Specifically, CCD cameras are fixed above and to the sides of the positions of the transformer bushing where the forged parts have been manufactured, images of the surfaces of the parts that have been produced are acquired using the CCD cameras, the acquired images are recorded as surface images, and the surface images are RGB images.
Because of factors such as environment and slight vibration of internal elements of a camera, noise inevitably appears on the acquired surface image, and in order to avoid the influence of the existence of the noise in the surface image on the result of subsequent analysis, gaussian filtering is used for respectively convolving each channel in the surface image, so that the surface image is denoised, and the precision and the quality of the surface image are improved. The gaussian filtering and denoising is a known technique, and the detailed description of the specific process is omitted.
And converting the surface image into a gray level image, and recording the gray level image as the surface gray level image. And performing edge detection on the surface gray level image by using a canny detection operator to obtain an edge image, wherein the edge image is a binary image and is recorded as a surface edge image. The canny detection operator is a known technology, and the scheme is not described in detail.
And according to the edge lines in the surface edge image, taking the areas divided by each closed edge line as the flaw to-be-detected areas, wherein the areas correspond to positions where the air tightness is possibly affected by the occurrence of cracks in the surface image, and further obtaining all the flaw to-be-detected areas in the surface image according to all the closed edge lines.
Step S002, obtaining the edge distance from any pixel point in the flaw to-be-detected area to the corresponding image edge of the surface image, and obtaining the azimuth gloss characteristic value of the flaw to-be-detected area by combining the corresponding value, the gray value and the corresponding edge distance of each pixel point in the flaw to-be-detected area in RGB three channels.
Specifically, it is known that the thermal cracking is easily caused near the edge of the casting and is characterized by an oxidation color and no metallic luster. The thermal cracks are cracks caused by the fact that the solid shrinkage of the casting is blocked when the casting is still in a state with low strength and plasticity at the last stage of solidification or shortly after solidification, and are often generated at corners, places with rapidly changed section thickness or places with slow local solidification of the casting, so that a defect region to be detected corresponding to the thermal cracks is also located at a position close to the edge of the casting.
Based on the position characteristics of the cracks, taking a flaw to-be-detected area as an example, taking any one pixel point in the flaw to-be-detected area as a target pixel point, respectively calculating the difference value between the Manhattan distance corresponding to the target pixel point and any one pixel point on the image edge corresponding to the surface image and the Chebyshev distance based on the coordinates of the target pixel point, and taking the minimum difference value as the edge distance from the target pixel point to the image edge corresponding to the surface image.
As an example, any pixel point A in the flaw detection area is taken, and the coordinate of the pixel point is recorded as
Wherein, in the step (A),
is the abscissa of the pixel point,
The vertical coordinate of the pixel point; and taking the distance difference between the corresponding Manhattan distance and the Chebyshev distance between the pixel point A and each pixel point on the edge of the surface image. It should be noted that the edge of the surface image refers to a peripheral boundary of the surface image, is not an edge of each region in the surface image, that is, an image edge of the surface image, and the shortest one of the distance differences is taken as an edge distance corresponding to the pixel point a
(ii) a Each pixel point in the flaw to-be-detected area can obtain a corresponding edge distance
。
The surface of the hot crack is in oxidation color, if the crack surface of the steel casting is approximately black, the aluminum alloy is dark gray and has no metallic luster, so that the gray value corresponding to each pixel point in the flaw detection area and the corresponding values under the RGB three channels are taken based on the color characteristic and are respectively marked as H, R, G and B. Because the gray value corresponding to the pixel point in the region corresponding to the thermal crack and the corresponding values under the three RGB channels are smaller, the azimuth gloss characteristic value of the defect region to be measured is obtained by combining the gray value corresponding to each pixel point in the defect region to be measured, the corresponding values under the three RGB channels and the corresponding edge distances, and the method for obtaining the azimuth gloss characteristic value of the defect region to be measured is as follows: arranging the edge distances of all pixel points in the flaw to-be-detected area from small to large to obtain a sequence, and acquiring pixel points corresponding to the first M edge distances in the sequence as first pixel points; m is a positive integer; calculating an average edge distance according to the edge distance of the first pixel point; the values corresponding to the RGB three channels are an R value, a B value and a G value, for each first pixel point, a first difference value between the maximum value and the R value, a second difference value between the maximum value and the B value and a third difference value between the maximum value and the G value are respectively calculated, the addition result of the first difference value, the second difference value and the third difference value is calculated, the addition result is used as a numerator, the gray value corresponding to the first pixel point is used as a denominator, and the corresponding ratio is obtained, and the ratios of all the first pixel points are added to obtain the ratio sum; and taking the product of the reciprocal of the average edge distance and the sum of the ratios as the azimuth gloss characteristic value of the flaw detected area.
As an example, the edge distances of each pixel point in the flaw detection area are arranged in a descending order to obtain a sequence, the pixel point corresponding to the first 50% edge distance in the sequence is obtained as a first pixel point, and a calculation formula of the azimuth gloss characteristic value of the flaw detection area is constructed according to the gray value corresponding to the first pixel point, the values corresponding to the RGB channels and the corresponding edge distances:
wherein, the first and the second end of the pipe are connected with each other,
the azimuth gloss characteristic value of the flaw to-be-detected area is obtained;
the edge distance of the ith first pixel point is calculated;
the number of the first pixel points;
the value of the ith first pixel point in the R channel is obtained;
the value of the ith first pixel point in a G channel is obtained;
the value of the ith first pixel point in the B channel is obtained;
the gray value of the ith first pixel point is obtained; and 255 is a maximum value.
It should be noted that the to-be-detected defective area corresponding to the thermal crack is located at a position close to the edge of the casting, so that the edge distance corresponding to each pixel point in the to-be-detected defective area is also small, when the crack grows larger, a partial area of the crack may extend to the middle of the casting, in order to avoid that the edge distance corresponding to the pixel points included in the partial areas is larger to influence the calculation value, the smaller edge distance in the edge distances corresponding to each pixel point in the to-be-detected defective area is selected for calculation, the thermal crack position corresponding to the to-be-detected defective area is represented by using the average value of the smaller edge distances, the larger the average value is, the more unlikely the to-be-detected defective area is to be a thermal crack, and the azimuth gloss characteristic value corresponding to the to-be-detected defective area is larger
The smaller; when the corresponding color of each pixel point in the flaw detection area is darker, the corresponding gray value is smaller, and the corresponding gray value is more acceptableCan be a hot crack pixel point, so that the azimuth gloss characteristic value of the flaw to-be-detected area
The larger, i.e., the more likely the defect region to correspond to the location of thermal cracks.
Step S003, performing linear fitting on all pixel points in the flaw detection area to obtain fitting goodness; carrying out corner point detection on the surface gray level image, and counting the number of corner points in a flaw to-be-detected area; thinning the flaw to-be-detected area to obtain a corresponding skeleton line, making a straight line for each pixel point on the skeleton line, and acquiring the crack width of the corresponding pixel point according to the intersection point of the straight line and the flaw to-be-detected area; and clustering the pixel points based on the crack width to obtain a first number of clusters, and combining the goodness of fit, the number of corner points, the crack width and the first number of clusters to obtain the hot crack tortuosity of the flaw detection area.
Specifically, the whole hot cracks are strip-shaped, germinate at the crystal boundary and expand along the crystal boundary, and are twisted in a zigzag manner and discontinuous, and the shapes of the hot cracks are uneven, so that the defect to-be-detected area is analyzed according to the characteristics.
Performing linear fitting on all pixel points contained in the flaw region to be measured by using a least square method to obtain corresponding goodness of fit
And the goodness of fit refers to the degree of fit between the position distribution of all pixel points contained in the flaw to-be-detected area and the fit straight line, and the greater the degree of fit, the more similar the flaw to-be-detected area and the strip are.
Because the hot cracks are zigzag, the edges of the corresponding to-be-detected flaw areas are also zigzag, and the number of the angular points is large, the angular point detection is carried out on the surface gray level image, and the number of the angular points contained in the to-be-detected flaw areas is counted and recorded as
。
The Hilditch algorithm is used for image thinning of the flaw detected region to obtain skeleton lines of the flaw detected region, as shown in fig. 2, wherein the Hilditch algorithm is a known technology, and is not described in detail herein. Numbering each pixel point on the skeleton line from one section of the skeleton line to the other end of the skeleton line in sequence, wherein for the ith pixel point, i =1, 2 and 3 \8230, N-1, N is the number of the pixel points on the skeleton line, and N is a positive integer; connecting the ith pixel point with the (i + 1) th pixel point to obtain a straight line as an ith straight line, making a target straight line perpendicular to the ith straight line by the ith pixel point to obtain two intersection points of the target straight line in the flaw to-be-detected area, and taking the Euclidean distance between the two intersection points as the crack width of the ith pixel point.
As an example, from one section of the skeleton line to the other end, the pixel points on the skeleton line are numbered one by one from the number 1, the pixel points corresponding to the numbers 1 and 2 are connected into a straight line as a first straight line corresponding to the pixel point of the number 1, the pixel point passing the number 1 is taken as a straight line perpendicular to the first straight line, and the straight line is recorded as a straight line
(ii) a Straight line
There will be two intersection points with the region to be measured for the flaw, and the euclidean distance between these two intersection points is calculated as the crack width C corresponding to the pixel point of number 1.
It should be noted that, because the crack width of each pixel point in the flaw detection area is obtained based on the straight line between adjacent pixel points, there is no crack width for the last pixel point on the skeleton line, that is, the crack width of the last pixel point on the skeleton line is not calculated in the present invention.
Because the hot cracks are twisted in a zigzag way, the widths of adjacent and near cracks corresponding to small parts with the same direction on the hot cracks are relatively close and correspond to the same cluster, the widths of the adjacent and near cracks at the position where the direction of the hot cracks changes correspond to a new cluster, the number of clusters of a flaw to-be-detected area corresponding to the hot cracks is large, a plane rectangular coordinate system is constructed by taking the crack width with the serial number as the abscissa and the pixel point with the serial number as the ordinate, and a plane rectangular coordinate system is constructed according to each coordinate point in the plane rectangular coordinate systemClustering all pixel points by using a DBSCAN algorithm to obtain a first number of clusters, which specifically comprises the following steps: obtaining a coordinate point of each pixel point on the plane rectangular coordinate system by taking the serial number of each pixel point on the skeleton line as the abscissa and the width of the crack corresponding to the pixel point as the ordinate
Using DBSCAN algorithm to the obtained pixel points corresponding to the coordinate points to obtain
And clustering, wherein 5 is the minimum number of points and 2 is the neighborhood radius in the DBSCAN algorithm.
Combining goodness of fit R, number of angular points
Crack width C and first number of clusters
And obtaining the hot crack tortuosity of the flaw to-be-detected area, specifically: respectively obtaining a crack width mean value, a maximum crack width and a minimum crack width based on all the crack widths; calculating a first difference absolute value of the maximum crack width and the minimum crack width; acquiring a difference absolute value between each crack width and the crack width mean value, and acquiring an average difference absolute value according to the difference absolute value corresponding to each crack width; and taking the product of the goodness of fit, the number of angular points, the first difference absolute value, the average difference absolute value and the first number of clusters as the hot crack tortuosity of the flaw detection area.
As an example, the hot crack tortuosity of the defect region to be detected is calculated by the formula:
wherein the content of the first and second substances,
is thermal crack tortuosity;
the number of crack widths;
is the ith crack width;
the mean value of the crack width is obtained;
maximum crack width;
is the minimum crack width;
the number of angular points in the flaw region to be measured;
a first number of clusters;
fitting goodness of a flaw to-be-detected area;
is the average absolute difference value.
It should be noted that the thermal cracks have uneven internal shape and thickness, and the difference between the maximum crack width and the minimum crack width is larger because the difference between the crack width at each pixel point on the skeleton line corresponding to the region to be detected of the defect is larger
The larger the heat cracking rate, the more likely the heat cracking rate is, and the larger the corresponding heat cracking tortuosity is; difference between crack width and average crack width
The larger, the average difference
The larger the defect area, the more the defect width is disordered, the more the defect area is possibly a thermal crack,the greater the corresponding thermal crack tortuosity; number of corner points in defect region to be measured
The more the number is, the more the edge of the defect region to be detected is bent, the more likely the edge is a thermal crack, and the higher the bending degree of the corresponding thermal crack is; first number of clusters
The more, the more the trend in the flaw detection area is zigzag and twisted, the more likely the flaw detection area is a thermal crack, and the larger the corresponding thermal crack tortuosity is; because the thermal cracks are strip-shaped, different cracks are not continuous, the greater the goodness of fit R of the corresponding to-be-detected flaw area is, the more similar the to-be-detected flaw area and the strip-shaped area are, and the greater the tortuosity of the corresponding thermal cracks is. The hot crack tortuosity measures the tortuosity and the width nonuniformity of the hot cracks corresponding to the region to be detected, the more tortuous and the more nonuniform the width, the larger the corresponding hot crack tortuosity, so that the hot crack tortuosity is in positive correlation with the goodness of fit of the region to be detected, the number of corner points of the region to be detected, the first difference absolute value between the maximum crack width and the minimum crack width, the average difference absolute value of the crack width and the first number of clusters.
S004, combining the azimuth gloss characteristic value and the hot crack tortuosity of the flaw to-be-detected area to obtain hot crack significance; and detecting the surface defects of the corresponding parts according to the hot crack significance of each defect region to be detected.
Specifically, the thermal cracks are easy to occur near the edge of the casting, are in the characteristic of oxidation color without metallic luster, and usually present the zigzag and uneven width characteristics, so the hot crack significance is obtained by combining the azimuth luster characteristic value and the thermal crack tortuosity of the to-be-detected-defect area, and each to-be-detected-defect area is evaluated by utilizing the thermal crack significance.
As an example, the azimuth gloss characteristic value of the flaw detection area is measured
And hot crack tortuosity
The product of (2) is taken as the hot cracking significance, then the hot cracking significance is
The calculation formula of (c) is:
。
it should be noted that the larger the azimuthal gloss characteristic value is, the more likely the defective region to be detected is thermal cracks, and the larger the thermal crack tortuosity is, the more likely the defective region to be detected is thermal cracks, so when the azimuthal gloss characteristic value corresponding to the defective region to be detected is
And hot crack tortuosity
The larger the heat check, the more likely it is, and the corresponding heat check is significant
The larger, i.e., both the azimuthal gloss characteristic and the thermal crack tortuosity correlate positively with thermal crack significance.
Acquiring the hot crack significance of each defective to-be-detected area by using the hot crack significance acquisition method, normalizing the hot crack significance of each defective to-be-detected area to obtain a corresponding hot crack significance normalized value, setting a normalized threshold value, and determining that the corresponding defective to-be-detected area is a hot crack when the hot crack significance normalized value is greater than or equal to the normalized threshold value; and setting a quantity threshold, counting the number of the detected areas with the defects confirmed as hot cracks, and confirming that the hot cracks exist on the surface of the corresponding part when the quantity is greater than or equal to the quantity threshold.
As an example, the normalization threshold value is set to be 0.6, when the hot crack significance normalization value corresponding to the detected defect area is greater than or equal to 0.6, the detected defect area is determined to be a hot crack, otherwise, the detected defect area is not a hot crack; and performing thermal crack confirmation on each defect to-be-detected area, counting the number of the defect to-be-detected areas which are confirmed to be thermal cracks, setting a number threshold value to be 1, confirming that the thermal crack defect exists on the surface of the corresponding part when one defect to-be-detected area is confirmed to be thermal cracks, and indicating that the airtightness is influenced by the thermal crack defect existing in the part.
In summary, in the embodiments of the present invention, surface images of each component of the transformer bushing are collected to obtain corresponding surface grayscale images; performing edge detection on the surface gray level image to obtain an area corresponding to a closed edge line as a flaw to-be-detected area; acquiring the edge distance from any pixel point in the flaw to-be-detected area to the corresponding image edge of the surface image, and acquiring the azimuth gloss characteristic value of the flaw to-be-detected area by combining the color characteristic of each pixel point in the flaw to-be-detected area and the corresponding edge distance; performing linear fitting on all pixel points in the flaw to-be-detected area to obtain goodness of fit; carrying out corner detection on the surface gray level image, and counting the number of corners in a flaw to-be-detected area; thinning the flaw to-be-detected area to obtain a corresponding skeleton line, making a straight line for each pixel point on the skeleton line, and acquiring the crack width of the corresponding pixel point according to the intersection point of the straight line and the flaw to-be-detected area; clustering the pixel points based on the crack width to obtain a first number of clusters, and combining the goodness of fit, the number of angular points, the crack width and the first number of clusters to obtain the hot crack tortuosity of the flaw detection area; combining the azimuth gloss characteristic value and the hot crack tortuosity of the flaw to-be-detected area to obtain the hot crack significance; and detecting the surface defects of the corresponding parts according to the hot crack significance of each defect region to be detected. The invention improves the accuracy and convenience of detecting the surface flaws of the corresponding parts according to the hot crack significance, and reduces the loss of manpower and resources.
Based on the same inventive concept as the method embodiment, the embodiment of the invention also provides a system for detecting surface flaws in transformer production, which comprises the following steps: the method comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to realize the steps in the embodiment of the method for detecting the surface defects in the production of the transformer, such as the steps shown in fig. 1. The method for detecting surface defects in transformer production has been described in detail in the above embodiments, and is not described again.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
All the embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from other embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that are within the spirit of the present invention are intended to be included therein.
Claims (9)
1. A surface flaw detection method in transformer production is characterized by comprising the following steps:
collecting surface images of all parts of a transformer bushing, wherein the surface images are RGB images, and obtaining surface gray level images corresponding to the RGB images; performing edge detection on the surface gray level image to obtain an area corresponding to a closed edge line as a flaw to-be-detected area;
acquiring the edge distance from any pixel point in the flaw to-be-detected area to the corresponding image edge of the surface image, and acquiring the azimuth gloss characteristic value of the flaw to-be-detected area by combining the corresponding value and gray value of each pixel point in the flaw to-be-detected area in RGB three channels and the corresponding edge distance;
performing linear fitting on all pixel points in the flaw to-be-detected area to obtain goodness of fit; carrying out corner point detection on the surface gray level image, and counting the number of corner points in a flaw to-be-detected area; thinning the flaw to-be-detected area to obtain a corresponding skeleton line, making a straight line for each pixel point on the skeleton line, and acquiring the crack width of the corresponding pixel point according to the intersection point of the straight line and the flaw to-be-detected area; clustering the pixel points based on the crack width to obtain a first number of clusters, and combining the goodness of fit, the number of angular points, the crack width and the first number of clusters to obtain the hot crack tortuosity of the flaw detection area;
combining the azimuth gloss characteristic value and the hot crack tortuosity of the flaw to-be-detected area to obtain the hot crack significance; and detecting the surface defects of the corresponding parts according to the hot crack significance of each defect region to be detected.
2. The method for detecting surface defects in transformer production according to claim 1, wherein the method for obtaining the edge distance from any pixel point in the region to be detected of the defects to the corresponding image edge of the surface image comprises:
and taking any pixel point in the region to be detected of the flaw as a target pixel point, respectively calculating the difference value between the corresponding Manhattan distance between the target pixel point and any pixel point on the image edge corresponding to the surface image and the Chebyshev distance based on the coordinates of the target pixel point, and taking the minimum difference value as the edge distance from the target pixel point to the image edge corresponding to the surface image.
3. The method as claimed in claim 1, wherein the method for obtaining the azimuth gloss characteristic value of the defect area to be tested by combining the corresponding values, gray values and corresponding edge distances of each pixel point in the defect area to be tested in three RGB channels comprises:
arranging the edge distances of all pixel points in the flaw to-be-detected area from small to large to obtain a sequence, and acquiring pixel points corresponding to the first M edge distances in the sequence as first pixel points; m is a positive integer;
calculating an average edge distance according to the edge distance of the first pixel point; the values corresponding to the RGB three channels are an R value, a B value and a G value, for each first pixel point, a first difference value between the maximum value and the R value, a second difference value between the maximum value and the B value and a third difference value between the maximum value and the G value are respectively calculated, the addition result of the first difference value, the second difference value and the third difference value is calculated, the addition result is used as a numerator, the gray value corresponding to the first pixel point is used as a denominator, and the corresponding ratio is obtained, and the ratios of all the first pixel points are added to obtain the ratio sum;
and taking the product of the reciprocal of the average edge distance and the sum of the ratios as the azimuth gloss characteristic value of the flaw detected area.
4. The method for detecting surface flaws in transformer production according to claim 1, wherein the method for drawing a straight line for each pixel point on the skeleton line and obtaining the crack width of the corresponding pixel point according to the intersection point of the straight line and the flaw detection area comprises:
numbering each pixel point on the skeleton line in sequence, wherein for the ith pixel point, i =1, 2 and 3 \ 8230n, N-1, N is the number of the pixel points on the skeleton line, and N is a positive integer; and connecting the ith pixel point and the (i + 1) th pixel point to obtain a straight line as an ith straight line, making a target straight line perpendicular to the ith straight line for the ith pixel point to obtain two intersection points of the target straight line in the flaw to-be-detected area, and taking the Euclidean distance between the two intersection points as the crack width of the ith pixel point.
5. The method for detecting surface flaws in transformer production according to claim 4, wherein the method for clustering pixel points based on crack widths to obtain a first number of clusters comprises:
and constructing a plane rectangular coordinate system by taking the serial number as an abscissa and the crack width of the pixel point corresponding to the serial number as an ordinate, and clustering all the pixel points by using a DBSCAN algorithm according to each coordinate point in the plane rectangular coordinate system to obtain a first number of clusters.
6. The method of claim 1, wherein the combining the goodness-of-fit, the number of corner points, the crack width, and the first number of clusters to obtain the hot crack tortuosity of the region under fault comprises:
respectively obtaining a crack width mean value, a maximum crack width and a minimum crack width based on all the crack widths; calculating a first difference absolute value of the maximum crack width and the minimum crack width; obtaining the absolute value of the difference value between the width of each crack and the mean value of the width of the crack, and obtaining an average absolute value of the difference value according to the absolute value of the difference value corresponding to the width of each crack;
and taking the product of the goodness of fit, the number of angular points, the first difference absolute value, the average difference absolute value and the first number of clusters as the hot crack tortuosity of the flaw detection area.
7. The method for detecting surface defects in transformer production according to claim 1, wherein the method for combining the azimuth gloss characteristic value and the hot crack tortuosity of the region to be detected of the defects to obtain the hot crack significance comprises the following steps:
and taking the product of the azimuth gloss characteristic value and the hot crack tortuosity of the flaw to-be-detected area as the hot crack significance.
8. The method for detecting surface defects in transformer production according to claim 1, wherein the method for detecting surface defects of corresponding parts according to the hot crack significance of each defect region to be detected comprises the following steps:
normalizing the hot crack significance of each defective to-be-detected area to obtain a corresponding hot crack significance normalized value, setting a normalized threshold value, and when the hot crack significance normalized value is larger than or equal to the normalized threshold value, determining that the corresponding defective to-be-detected area is a hot crack;
and setting a quantity threshold, counting the number of the detected areas with the defects confirmed as hot cracks, and confirming that the hot cracks exist on the surface of the corresponding part when the quantity is greater than or equal to the quantity threshold.
9. A system for detecting surface flaws in transformer production, the system comprising: a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the steps of the method for detecting surface defects in transformer production according to any one of the preceding claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310251600.4A CN115965623B (en) | 2023-03-16 | 2023-03-16 | Surface flaw detection method and system in transformer production |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310251600.4A CN115965623B (en) | 2023-03-16 | 2023-03-16 | Surface flaw detection method and system in transformer production |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115965623A true CN115965623A (en) | 2023-04-14 |
| CN115965623B CN115965623B (en) | 2023-05-30 |
Family
ID=85899689
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310251600.4A Active CN115965623B (en) | 2023-03-16 | 2023-03-16 | Surface flaw detection method and system in transformer production |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115965623B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116258713A (en) * | 2023-05-11 | 2023-06-13 | 青岛穗禾信达金属制品有限公司 | Welding processing detection method for metal cabinet |
| CN116309550A (en) * | 2023-05-11 | 2023-06-23 | 聊城市飓风工业设计有限公司 | Integrated circuit patch abnormality identification method based on image processing |
| CN116385434A (en) * | 2023-06-02 | 2023-07-04 | 同济检测(济宁)有限公司 | Intelligent detection method for precast beam cracks |
| CN117237646A (en) * | 2023-11-15 | 2023-12-15 | 深圳市润海电子有限公司 | PET high-temperature flame-retardant adhesive tape flaw extraction method and system based on image segmentation |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130135460A1 (en) * | 2011-05-27 | 2013-05-30 | Airbus Operations Gmbh | Method for the detection of a possible joint defect in a friction stir weld seam |
| CN105957088A (en) * | 2016-05-12 | 2016-09-21 | 广东电网有限责任公司云浮供电局 | Method and system for monitoring transformer composite insulator sleeve based on computer vision |
| CN115063423A (en) * | 2022-08-18 | 2022-09-16 | 启东市嘉信精密机械有限公司 | Mechanical casting cold and hot crack self-adaptive identification method based on computer vision |
| CN115222733A (en) * | 2022-09-19 | 2022-10-21 | 山东金源不锈钢制品有限公司 | Metal component surface defect detection method based on image recognition |
-
2023
- 2023-03-16 CN CN202310251600.4A patent/CN115965623B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130135460A1 (en) * | 2011-05-27 | 2013-05-30 | Airbus Operations Gmbh | Method for the detection of a possible joint defect in a friction stir weld seam |
| CN105957088A (en) * | 2016-05-12 | 2016-09-21 | 广东电网有限责任公司云浮供电局 | Method and system for monitoring transformer composite insulator sleeve based on computer vision |
| CN115063423A (en) * | 2022-08-18 | 2022-09-16 | 启东市嘉信精密机械有限公司 | Mechanical casting cold and hot crack self-adaptive identification method based on computer vision |
| CN115222733A (en) * | 2022-09-19 | 2022-10-21 | 山东金源不锈钢制品有限公司 | Metal component surface defect detection method based on image recognition |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116258713A (en) * | 2023-05-11 | 2023-06-13 | 青岛穗禾信达金属制品有限公司 | Welding processing detection method for metal cabinet |
| CN116309550A (en) * | 2023-05-11 | 2023-06-23 | 聊城市飓风工业设计有限公司 | Integrated circuit patch abnormality identification method based on image processing |
| CN116258713B (en) * | 2023-05-11 | 2023-07-21 | 青岛穗禾信达金属制品有限公司 | Welding processing detection method for metal cabinet |
| CN116309550B (en) * | 2023-05-11 | 2023-08-04 | 聊城市飓风工业设计有限公司 | Integrated circuit patch abnormality identification method based on image processing |
| CN116385434A (en) * | 2023-06-02 | 2023-07-04 | 同济检测(济宁)有限公司 | Intelligent detection method for precast beam cracks |
| CN116385434B (en) * | 2023-06-02 | 2023-08-08 | 同济检测(济宁)有限公司 | Intelligent detection method for precast beam cracks |
| CN117237646A (en) * | 2023-11-15 | 2023-12-15 | 深圳市润海电子有限公司 | PET high-temperature flame-retardant adhesive tape flaw extraction method and system based on image segmentation |
| CN117237646B (en) * | 2023-11-15 | 2024-01-30 | 深圳市润海电子有限公司 | PET high-temperature flame-retardant adhesive tape flaw extraction method and system based on image segmentation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115965623B (en) | 2023-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115965623A (en) | Surface flaw detection method and system in transformer production | |
| CN114882044B (en) | Metal pipe surface quality detection method | |
| CN115082464A (en) | Method and system for identifying welding seam data in welding process of dust remover | |
| CN116091504B (en) | Connecting pipe connector quality detection method based on image processing | |
| CN115063423B (en) | Self-adaptive identification method for cold and hot cracks of mechanical castings based on computer vision | |
| CN115841434A (en) | Infrared image enhancement method for gas concentration analysis | |
| CN116721107B (en) | Intelligent monitoring system for cable production quality | |
| CN115937216B (en) | Magnetic rotor appearance quality detection method for new energy automobile | |
| CN111667470B (en) | Industrial pipeline flaw detection inner wall detection method based on digital image | |
| CN115063430B (en) | Electric pipeline crack detection method based on image processing | |
| CN115690108A (en) | Aluminum alloy rod production quality evaluation method based on image processing | |
| CN115330796B (en) | Copper wire tinning defect identification method | |
| CN113155839A (en) | Steel plate outer surface defect online detection method based on machine vision | |
| CN115272336A (en) | Metal part defect accurate detection method based on gradient vector | |
| CN112669295A (en) | Lithium battery pole piece defect detection method based on secondary threshold segmentation theory | |
| CN115239738A (en) | Intelligent detection method for automobile part configuration | |
| CN116168037A (en) | Method and system for calculating bending degree of wire crimping based on image processing | |
| CN107230212B (en) | Vision-based mobile phone size measuring method and system | |
| CN111638218A (en) | Method for detecting surface defects of coating | |
| CN114881961A (en) | Glass fiber plate crack detection method and device | |
| CN117152129B (en) | Visual detection method and system for surface defects of battery cover plate | |
| CN117351017A (en) | Visual defect detection method for photovoltaic steel structure component | |
| CN116612112B (en) | Visual inspection method for surface defects of bucket | |
| CN116862910A (en) | Visual detection method based on automatic cutting production | |
| CN116664567A (en) | Solid insulation switch cabinet quality assessment method and system |
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 | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 518000 room 1101-1105, building 6, Hengda fashion Huigu building (East District), Fulong Road, Henglang community, Dalang street, Longhua District, Shenzhen, Guangdong Patentee after: Cenke Technology (Shenzhen) Group Co.,Ltd. Address before: Room 1101-1105, Building 6, Evergrande Fashion Huigu Building (East Area), Fulong Road, Henglang Community, Dalang Street, Longhua District, Shenzhen, Guangdong 518110 Patentee before: SHENZHEN CENKER ENTERPRISE Ltd. |