CN110909751B - Visual identification method, system and medium for transformer substation insulator cleaning robot - Google Patents
Visual identification method, system and medium for transformer substation insulator cleaning robot Download PDFInfo
- Publication number
- CN110909751B CN110909751B CN201911175633.5A CN201911175633A CN110909751B CN 110909751 B CN110909751 B CN 110909751B CN 201911175633 A CN201911175633 A CN 201911175633A CN 110909751 B CN110909751 B CN 110909751B
- Authority
- CN
- China
- Prior art keywords
- insulator
- curve
- ellipse
- point
- cleaning robot
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention discloses a visual identification method, a system and a medium for a transformer substation insulator cleaning robot, wherein the step of identifying through an insulator umbrella skirt edge comprises the steps of extracting all edges in an insulator image, removing straight line segments and leaving curve segments; calculating the concavity and convexity of all the curve segments, and only reserving the curve segments only with concavity and convexity; judging whether the curve segment is a part of an ellipse or not for each curve segment, if so, retaining the curve segment and recording the corresponding elliptic secondary curve coordinate of the curve segment, and performing duplication removal for the elliptic secondary curve coordinate; the method can realize detection and identification of the umbrella skirt edge of the insulator, so that the insulator cleaning robot can climb on the insulator, the cleaning robot can perform targeted descaling, the descaling efficiency is improved, and the method has the advantages of high identification accuracy and capability of efficiently, accurately and quickly identifying the insulator.
Description
Technical Field
The invention relates to the field of live-wire work automation of power systems, in particular to a visual identification method, a system and a medium for a transformer substation insulator cleaning robot.
Background
With the development of science and technology, various robots of a transformer substation are widely applied, and the functions of the robots are more and more powerful. Stable and reliable operation, and simultaneously, the bearing capacity to extreme severe natural conditions is required to be better. The aging of the insulator and the gathering of surface pollutants can cause the resistance value of the insulator to be reduced, and further cause the deterioration and the breakdown of the insulator, thereby causing major safety accidents. The safety and reliability of the insulator are guaranteed, so that the stable operation of a power line is guaranteed, and the method has important significance for guaranteeing the national power safety. The traditional cleaning mode of the insulator mainly comprises manual scrubbing and hand-held high-pressure water cannon washing, and the manual cleaning operation has the problems of high labor intensity, low washing efficiency, incapability of ensuring the washing effect and the like, and simultaneously has certain operation potential safety hazards. The insulator cleaning robot for the transformer substation can effectively reduce the manual operation intensity, reduce the probability of operation safety accidents and improve the cleaning effect, has very important significance for routine maintenance of the transformer substation, and realizes accurate identification, positioning and pollution degree judgment of the insulator.
Disclosure of Invention
The technical problems to be solved by the invention are as follows: aiming at the problems in the prior art, the invention provides a visual identification method, a system and a medium for a transformer substation insulator cleaning robot, wherein the ellipse of the insulator umbrella skirt edge can be identified based on an insulator image through the insulator umbrella skirt edge identification step, and the detection and identification of the insulator umbrella skirt edge are realized by fully utilizing the characteristics of the insulator umbrella skirt edge, so that the insulator cleaning robot can climb on the insulator, the cleaning robot is favorable for performing targeted descaling and the descaling efficiency is improved.
In order to solve the technical problems, the invention adopts the technical scheme that:
a visual identification method for a transformer substation insulator cleaning robot comprises the step of identifying an umbrella skirt edge of an insulator, and the detailed steps comprise:
1) extracting all edges in the insulator image, removing straight line segments and leaving curve segments;
2) calculating the concavity and convexity of all the curve segments, and only reserving the curve segments only with concavity and convexity;
3) judging whether the curve segment is a part of an ellipse or not for each curve segment, if so, retaining the curve segment and recording the corresponding elliptic secondary curve coordinate of the curve segment, and performing duplication removal for the elliptic secondary curve coordinate;
4) and clustering all the ellipses by using a clustering algorithm to obtain the ellipses of the umbrella skirt edges of the insulators.
Optionally, the detailed steps of step 1) include:
1.1) carrying out binarization processing on the insulator image and extracting a skeleton of the insulator image;
1.2) aiming at the extracted skeleton, searching a first white point in the image by traversing, then, starting from the point to extend to two sides to search for adjacent points, storing the point and then putting the point black when finding one point, and when finding a point without the adjacent point, indicating that the end search is finished and determining one end of the line segment; finally finding out all straight line segments and curve segments in the skeleton;
1.3) removing the stored straight line segments by adopting Hough straight line detection to leave curve segments.
Optionally, the detailed steps of step 2) include:
2.1) calculating the concavity and convexity f (x, y) of all curve segments according to the following formula;
in the above formula, f (x, y) is the calculation result of the concave-convex property, (x, y) is the horizontal and vertical coordinates of a certain point on the edge line, the subscript L represents the left end point of the edge line coordinate, and R represents the right end point of the edge line coordinate;
2.2) aiming at the roughness f (x, y) of each curve segment, if the roughness f (x, y) is constantly larger than zero, the curve segment is convex, the curve segment is kept, if the roughness f (x, y) is constantly smaller than zero, the curve segment is concave, and if the roughness f (x, y) has a sign difference result, the curve segment is judged not to meet the requirement and is deleted.
Optionally, the detailed steps of step 3) include:
3.1) constructing an edge point set D from the curve segment, wherein the value of an initialization counter C is 0, and a cycle count k is 0;
3.2) randomly selecting 5 points p 1-p 5 from the edge point set D;
3.3) determination of the quadratic curve Ax by means of the 5 points p1 to p5 2 +Bxy+Cy 2 A parameter a, B, C, D, E where x, y denote the coordinates of the point, if and only if B is satisfied, where + Dx + Ey +1 ═ 0 2 -4AC>When 0, judging the quadratic curve is an ellipse, executing the step 3.4), otherwise, skipping to execute the step 3.5);
3.4) substitution of the point p6 into the quadratic curve Ax defined by the points p1 to p5 from the point p6 randomly in the edge point set D 2 +Bxy+Cy 2 + Dx + Ey +1 ═ 0 gives d6 ═ Ax 6 2 +Bx 6 y 6 +cy 6 2 +Dx 6 +Ey 6 +1|, where d6 denotes the absolute value of the distance of P6 from the ellipse P, x 6 ,y 6 Coordinates representing point p 6;
3.5) traversing all edge points in the edge point set D, calculating the distance D between the edge point and a possible ellipse, adding 1 to a counter C if D is less than or equal to an allowable error delta, removing the edge point from the edge point set D, and storing the edge point into a possible ellipse edge point set P; otherwise, skipping to execute the step 3.6); if the value of the counter C is greater than a threshold value T g If the possible ellipse is a real ellipse, outputting a quadratic curve and a central point coordinate, otherwise, exiting;
3.6) adding 1 to the loop count k, if the loop count k is larger than the specified maximum loop time Kmax, ending, otherwise, jumping to execute the step 3.2).
Optionally, the step 4) of clustering all ellipses by using a clustering algorithm to obtain the ellipse of the insulator umbrella skirt specifically means that the center coordinates of the ellipses are clustered by using the clustering algorithm, and the ellipse of the cluster with the smallest difference value between the center Y coordinate of the ellipse and the Y coordinate of the insulator is reserved as the ellipse of the insulator umbrella skirt obtained by clustering according to the clustering result.
Optionally, after the step 4), calculating a ratio of a long side to a short side of the ellipse of each insulator umbrella skirt edge, and selecting the flattest ellipse as the ellipse of the target insulator umbrella skirt edge finally identified.
Optionally, the method further comprises a step of classifying insulator contamination, and the detailed steps include:
s1) collecting an insulator image;
s2) filtering the background interference and noise of the insulator image and dividing the image into insulator sub-images;
s3) respectively converting the RGB images of the insulator sub-images into an LAB color model and an HSI color model, and taking the mean value as a color characteristic value;
s4) inputting the color characteristic value into a machine learning classification model which is trained in advance to obtain insulator pollution classification results of each insulator, wherein the machine learning classification model is trained in advance to establish mapping between the color characteristic value of the insulator and the insulator pollution classification results.
Optionally, the machine learning classification model is an SVM classifier, and step S4) is preceded by a step of performing parameter optimization on the SVM classifier by using a genetic algorithm to determine a penalty parameter of the RBF kernel function c and a penalty parameter of the g kernel function, so that the sample training data which is linearly inseparable in the low-dimensional space is mapped to the high-dimensional feature space by the RBF kernel function c to become linearly separable.
Furthermore, the invention also provides a visual identification system for the transformer substation insulator cleaning robot, which comprises a computer device programmed or configured to execute the steps of the visual identification method for the transformer substation insulator cleaning robot, or a computer program programmed or configured to execute the visual identification method for the transformer substation insulator cleaning robot is stored on a memory of the computer device.
Furthermore, the present invention also provides a computer-readable storage medium having stored thereon a computer program programmed or configured to execute the visual recognition method for a substation insulator cleaning robot.
Compared with the prior art, the invention has the following advantages:
1. the step of identifying the umbrella skirt edge of the insulator by the vision identification method for the insulator cleaning robot of the transformer substation can identify the ellipse of the umbrella skirt edge of the insulator based on the insulator image, and the detection and identification of the umbrella skirt edge of the insulator are realized by fully utilizing the characteristics of the umbrella skirt edge of the insulator, so that the insulator cleaning robot can climb on the insulator, the cleaning robot is favorable for performing targeted descaling, and the descaling efficiency is improved.
2. The step of identifying the skirt edge of the insulator by using the visual identification method for the insulator cleaning robot of the transformer substation comprises the steps of identifying the edge, removing a straight line segment and leaving a curve segment; calculating the concavity and convexity of all the curve segments, and only reserving the curve segments only with concavity and convexity; judging whether the curve segment is a part of an ellipse or not for each curve segment, if so, retaining the curve segment and recording the corresponding elliptic secondary curve coordinate of the curve segment, and performing duplication removal for the elliptic secondary curve coordinate; the method has the advantages that the method can accurately identify the ellipse of the umbrella skirt edge of the insulator, has the advantage of high identification accuracy, can efficiently, accurately and rapidly identify the insulator, enables the insulator cleaning robot to climb on the insulator, improves the working efficiency and achieves the purpose of automatically cleaning the insulator.
Drawings
Fig. 1 is a basic flow chart of the identification of the skirt of the umbrella insulator in the embodiment of the invention.
Fig. 2 is a schematic diagram of a substation insulator umbrella skirt profile at different viewing angles.
Fig. 3 is a climbing schematic diagram of the insulator cleaning robot of the transformer substation in the embodiment of the invention.
Fig. 4 is a schematic flow chart of insulator contamination classification performed in the embodiment of the present invention.
FIG. 5 is a schematic diagram illustrating a process of training and using a machine learning classification model according to an embodiment of the present invention.
Detailed Description
As shown in fig. 1, the visual identification method for the transformer substation insulator cleaning robot in the embodiment includes a step of identifying an insulator umbrella skirt edge, and includes the following detailed steps:
1) extracting all edges in the insulator image, removing straight line segments and leaving curve segments;
2) calculating the concavity and convexity of all the curve segments, and only reserving the curve segments only with concavity and convexity;
3) judging whether the curve segment is a part of an ellipse or not for each curve segment, if so, retaining the curve segment and recording the corresponding elliptic secondary curve coordinate of the curve segment, and performing duplication removal for the elliptic secondary curve coordinate;
4) and clustering all the ellipses by using a clustering algorithm to obtain the ellipses of the umbrella skirt edges of the insulators.
Referring to fig. 1, similar to a general image recognition method, in order to improve the accuracy of image recognition, step 1) of this embodiment further includes a step of preprocessing an insulator image, and processing manners such as filtering, noise reduction, and graying can be selected as needed.
In this embodiment, the detailed steps of step 1) include:
1.1) carrying out binarization processing on the insulator image and extracting a skeleton of the insulator image;
1.2) aiming at the extracted skeleton, searching a first white point in the image by traversing, then, starting from the point to extend to two sides to search for adjacent points, storing the point and then putting the point black when finding one point, and when finding a point without the adjacent point, indicating that the end search is finished and determining one end of the line segment; finally finding out all straight line segments and curve segments in the skeleton;
1.3) removing the stored straight line segments by adopting Hough straight line detection to leave curve segments.
In this embodiment, the detailed steps of step 2) include:
2.1) calculating the concavity and convexity f (x, y) of all curve segments according to the following formula;
in the above formula, f (x, y) is the calculation result of the unevenness, and (x, y) is the horizontal and vertical coordinates of a certain point on the edge line; x is the number of L And x R Respectively represent edge linesLeft and right end points of abscissa, y L And y R Respectively representing the left end point and the right end point of the edge line vertical coordinate; the subscript L represents the left end point of the edge line coordinates and R represents the right end point of the edge line coordinates.
2.2) aiming at the roughness f (x, y) of each curve segment, if the roughness f (x, y) is constantly larger than zero, the curve segment is convex, if the roughness f (x, y) is constantly smaller than zero, the curve segment is concave, the curve segment is kept, and if the roughness f (x, y) has a different sign result, the curve segment is judged to be not satisfied with the requirement and is deleted.
In this embodiment, the detailed steps of step 3) include:
3.1) constructing an edge point set D from the curve segment, initializing the value of a counter C to be 0, and setting a cycle count k to be 0;
3.2) randomly selecting 5 points p 1-p 5 from the edge point set D;
3.3) determination of the quadratic curve Ax by means of the 5 points p1 to p5 2 +Bxy+Cy 2 A parameter a, B, C, D, E where x, y denote the coordinates of the point, if and only if B is satisfied, where + Dx + Ey +1 ═ 0 2 -4AC>When 0, judging the quadratic curve is an ellipse, executing the step 3.4), otherwise, skipping to execute the step 3.5);
3.4) substitution of the point p6 into the quadratic curve Ax defined by the points p1 to p5 from the point p6 randomly in the edge point set D 2 +Bxy+Cy 2 + Dx + Ey +1 ═ 0 gives d6 ═ Ax 6 2 +Bx 6 y 6 +cy 6 2 +Dx 6 +Ey 6 +1|, where d6 denotes the absolute value of the distance of P6 from the ellipse P, x 6 ,y 6 Coordinates representing point p 6;
3.5) traversing all edge points in the edge point set D, calculating the distance D between the edge point and a possible ellipse, if D is less than or equal to an allowable error delta, adding 1 to a counter C, removing the edge point from the edge point set D, and storing the edge point into a possible ellipse edge point set P; otherwise, skipping to execute the step 3.6); if the value of the counter C is greater than a threshold value T g If the possible ellipse is a real ellipse, outputting a quadratic curve and a central point coordinate, otherwise, exiting;
3.6) adding 1 to the loop count k, if the loop count k is larger than the specified maximum loop time Kmax, ending, otherwise, skipping to execute the step 3.2).
In this embodiment, the step 4) of clustering all ellipses by using a clustering algorithm to obtain the ellipse of the insulator umbrella skirt specifically means that the center coordinates of the ellipses are clustered by using the clustering algorithm, and the ellipse of the cluster with the smallest difference between the center Y coordinate of the ellipse and the center Y coordinate of the insulator is reserved as the ellipse of the insulator umbrella skirt obtained by clustering according to the clustering result.
In this embodiment, after the step 4), the method further includes calculating a ratio of a long side to a short side of the ellipse of each insulator umbrella skirt, and selecting the flattest ellipse as the ellipse of the target insulator umbrella skirt obtained through final recognition, so that the cleaning robot stays at the flattest ellipse to clean dirt between the umbrella skirts. As shown in fig. 3, when the transformer substation insulator cleaning robot climbs over the insulator, the supporting part is the edge position of the umbrella skirt edge of the insulator, so that according to the ratio of the long edge to the short edge of the ellipse, the flattest ellipse is selected as the ellipse of the target insulator umbrella skirt edge obtained by final recognition, so that the cleaning robot stays at the flattest ellipse to clean dirt between the umbrella skirt edges of the next insulator.
In addition, in order to implement cleaning of insulators with different pollution levels to improve cleaning efficiency, the visual identification method for the transformer substation insulator cleaning robot in this embodiment further includes a step of classifying pollution of the insulators, as shown in fig. 4 and 5, the detailed steps include:
s1) collecting an insulator image;
s2) filtering background interference and noise points of the insulator image and dividing the image into insulator sub-images;
s3) respectively converting the RGB images of the insulator subgraphs into an LAB color model and an HSI color model, and taking the mean value as a color characteristic value;
s4) inputting the color characteristic value into a machine learning classification model which is trained in advance to obtain insulator pollution classification results of each insulator, wherein the machine learning classification model is trained in advance to establish mapping between the color characteristic value of each insulator and the insulator pollution classification results.
In the embodiment, the average value of the color components is provided to express the color characteristics through the stronger color difference resolution capability of the LAB and the stronger visual perception capability of the HSI, and finally, a classification model is constructed through an SVM (support vector machine) support vector machine to classify the insulator contamination, so that the classification accuracy is improved. When the RGB image of each insulator subgraph is converted into an LAB color model respectively, the RGB color space can not be directly converted into an LAB color space, the RGB color space is converted into an XYZ color space by means of the XYZ color space, and then the XYZ color space is converted into the Lab color space, namely: RGB- > XYZ, and then XYZ- > LAB. Since the color space conversion method is a conventional method, the detailed steps will not be described herein. The color space conversion method for converting the RGB images of the insulator subgraphs into HSI color models is a conventional known method, so detailed steps are not explained herein;
in this embodiment, the machine learning classification model is an SVM classifier, and referring to fig. 5, step S4) further includes a step of performing parameter optimization on the SVM classifier by using a genetic algorithm to determine a penalty parameter of the RBF kernel function c and a g kernel function parameter, so that the sample training data which is linearly inseparable in the low-dimensional space is mapped to the high-dimensional feature space through the RBF kernel function c to become linearly separable. As an optional implementation manner, in the embodiment, the insulator contamination classification results of the machine learning classification model which is trained in advance are totally divided into five grades, so that image information of 5 insulators with different contamination grades needs to be collected during training, images of the insulators are respectively calibrated, background interference and noise caused by dust and the like are filtered by mean filtering, and image segmentation is performed. Referring to fig. 5, the detailed training steps of the SVM classifier include:
firstly, constructing a training set and a testing set by collecting insulator images, and respectively attaching labels to the training set and the testing set;
and step two, performing parameter optimization on the SVM classifier by using the training set and the label thereof and adopting a genetic algorithm to determine the penalty parameter of the RBF kernel function c and the g kernel function parameter, so that the sample training data which is linear and inseparable in the low-dimensional space is mapped to the high-dimensional feature space through the RBF kernel function c to become linearly separable. The method comprises the following steps of performing parameter optimization on an SVM classifier by adopting a genetic algorithm to determine an RBF kernel function c penalty parameter and a g kernel function parameter: (1) preparing sample data; (2) initializing internal parameters of a genetic algorithm; (3) generating parameter pairs, and searching parameters with the best fitness on a training set by using cross validation; (4) and (5) retraining by using the optimal parameters to obtain a support vector machine prediction model.
And step three, training the SVM classifier by using a training set and labels thereof according to the optimally determined penalized parameters of the RBF kernel function c and the g kernel function parameters, and then completing the training of the SVM classifier by using the standard rate of the test result of the test set if the standard rate meets the requirement, otherwise continuing to execute the step one to train the SVM classifier.
In summary, the visual identification method for the transformer substation insulator cleaning robot enables the transformer substation insulator cleaning to automatically complete the transformer substation insulator pollution classification identification, the insulator umbrella skirt edge is automatically positioned through vision, the input of manpower and material resources is reduced, and the working efficiency of the transformer substation insulator cleaning is improved. When the insulator cleaning robot of the transformer substation climbs the insulator, dirt between the insulator umbrella skirt edges is difficult to clean, and gullies are formed in the lower portions of some types of the insulator umbrella skirt edges, so that the insulator umbrella skirt edges can be identified, the insulator cleaning robot of the transformer substation can perform targeted descaling, and the descaling efficiency is improved. The embodiment is that the image of insulator in the transformer substation is gathered through the camera in real time, because the insulator of transformer substation cleans the robot and need climb to insulator chain top behind the upper boom to accomplish and clean, detect the operation task at the climbing in-process. The camera of the insulator cleaning robot of the transformer substation collects and processes image information, when a new insulator shed is detected to appear and is located in the middle position of an image, the insulator shed is regarded as reaching a grabbing position, and a clamping mechanism of the insulator cleaning robot of the transformer substation clamps the recognized shed. Because the dirt between the insulating umbrella skirt edges is difficult to clean, and gullies are formed in the lower parts of some types of insulating umbrella skirt edges, the insulating umbrella skirt edges are identified, so that the cleaning robot can conveniently perform targeted descaling, and the descaling efficiency is improved. And when the climbing movement is completed once, the collected image sequence is analyzed, the color characteristic value is extracted, and then the color characteristic value is input into a machine learning classification model trained before to obtain the pollution grade of the insulator, and the cleaning mode is automatically selected by the insulator cleaning robot. The cleaning mechanism cleans the lower surface of the upper insulator and the upper surface of the lower insulator. The camera that cleans the mechanism and carry can gather insulator surface's image information when the mechanism gyration for detect the insulator state. After the insulator is cleaned in the automatically selected cleaning mode, the insulator pollution grade algorithm is needed to detect the insulator, and if the insulator is polluted, the cleaning mode is automatically selected for cleaning. When the cleaning frequency reaches 3 times and the cleaning effect is still not good, the control center is informed to carry out manual inspection. When the transformer substation insulator cleaning robot cleans from bottom to top, the outline of the umbrella skirt edge of the insulator, which is shot by the camera of the transformer substation insulator cleaning robot, is similar to an ellipse under different observation angles. As shown in fig. 3, the ellipse that approximates the umbrella skirt of the insulator closest to the camera is flattest and rounder the farther away. Therefore, the step of identifying the umbrella skirt edge of the insulator in the visual identification method for the insulator cleaning robot of the transformer substation can accurately identify the ellipse of the umbrella skirt edge of the insulator, has the advantage of high identification accuracy, and can efficiently, accurately and rapidly identify the insulator, so that the insulator cleaning robot can climb on the insulator, the working efficiency is improved, and the purpose of automatically cleaning the insulator is achieved.
In addition, the present embodiment also provides a visual recognition system for a substation insulator cleaning robot, which includes a computer device programmed or configured to execute the steps of the visual recognition method for the substation insulator cleaning robot, or a computer program programmed or configured to execute the visual recognition method for the substation insulator cleaning robot is stored in a memory of the computer device. Furthermore, the present embodiment also provides a computer-readable storage medium having stored thereon a computer program programmed or configured to execute the visual recognition method for the substation insulator cleaning robot.
The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may occur to those skilled in the art without departing from the principle of the invention, and are considered to be within the scope of the invention.
Claims (8)
1. The visual identification method for the transformer substation insulator cleaning robot is characterized by comprising the step of identifying an insulator umbrella skirt edge, and the detailed steps comprise:
1) extracting all edges in the insulator image, removing straight line segments and leaving curve segments;
2) calculating the concavity and convexity of all the curve segments, and only reserving the curve segments only with concavity and convexity;
3) judging whether each curve segment is a part of an ellipse or not, if so, retaining the curve segment and recording the corresponding ellipse secondary curve coordinate of the curve segment, and performing duplication elimination according to the secondary curve coordinate of the recorded ellipse;
4) clustering all the ellipses by using a clustering algorithm to obtain the ellipses of the umbrella skirt edges of the insulators;
the detailed steps of the step 1) comprise:
1.1) carrying out binarization processing on the insulator image and extracting a skeleton of the insulator image;
1.2) aiming at the extracted skeleton, searching a first white point in the image by traversing, then, starting from the point to extend to two sides to search for adjacent points, storing the point and then putting the point black when finding one point, and when finding a point without the adjacent point, indicating that the end search is finished and determining one end of the line segment; finally finding out all straight line segments and curve segments in the skeleton;
1.3) removing the stored straight line segments by adopting Hough straight line detection to leave curve segments;
the detailed steps of the step 2) comprise:
2.1) calculating the concavity and convexity f (x, y) of all curve segments according to the following formula;
in the above formula, f (x, y) is the calculation result of the concavity and convexity, where (x, y) is the horizontal and vertical coordinates of a certain point on the edge line, the subscript L represents the left end point of the edge line coordinate, and R represents the right end point of the edge line coordinate;
2.2) aiming at the roughness f (x, y) of each curve segment, if the roughness f (x, y) is constantly larger than zero, the curve segment is convex, if the roughness f (x, y) is constantly smaller than zero, the curve segment is concave, the curve segment is kept, and if the roughness f (x, y) has a different sign result, the curve segment is judged to be not satisfied with the requirement and is deleted.
2. The visual identification method for the insulator cleaning robot of the transformer substation according to claim 1, wherein the detailed steps of the step 3) comprise:
3.1) constructing an edge point set D from the curve segment, initializing the value of a counter C to be 0, and setting a cycle count k to be 0;
3.2) randomly selecting 5 points p 1-p 5 from the edge point set D;
3.3) determination of the quadratic curve Ax by means of the 5 points p1 to p5 2 +Bxy+Cy 2 A parameter a, B, C, D, E where x, y denote the coordinates of the point, if and only if B is satisfied, where + Dx + Ey +1 ═ 0 2 -4AC > 0, determining the quadratic curve is an ellipse, performing step 3.4), otherwise, skipping to perform step 3.5);
3.4) substitution of the point p6 into the quadratic curve Ax defined by the points p1 to p5 from the point p6 randomly in the edge point set D 2 +Bxy+Cy 2 + Dx + Ey +1 ═ 0 gives d6 ═ Ax 6 2 +Bx 6 y 6 +cy 6 2 +Dx 6 +Ey 6 +1|, where d6 denotes the absolute value of the distance of P6 from the ellipse P, x 6 ,y 6 Coordinates representing point p 6;
3.5) traversing all edge points in the edge point set DCalculating the distance D between the edge point and the possible ellipse, if D is less than or equal to the tolerance delta, adding 1 to the counter C, removing the edge point from the edge point set D, and storing the edge point into the possible ellipse edge point set P; otherwise, skipping to execute the step 3.6); if the value of the counter C is greater than a threshold value T g If the possible ellipse is a real ellipse, outputting a quadratic curve and a central point coordinate, otherwise, exiting;
3.6) adding 1 to the loop count k, if the loop count k is larger than the specified maximum loop time Kmax, ending, otherwise, skipping to execute the step 3.2).
3. The visual identification method for the insulator cleaning robot of the substation according to claim 1, wherein the step 4) of clustering all ellipses by using a clustering algorithm to obtain the ellipse of the insulator umbrella skirt specifically means that the center coordinates of the ellipses are clustered by using the clustering algorithm, and the ellipse with the smallest difference value between the center Y coordinate of the ellipse and the Y coordinate of the insulator is reserved as the ellipse of the insulator umbrella skirt obtained by clustering according to the clustering result.
4. The visual identification method for the transformer substation insulator cleaning robot according to claim 1, wherein the step 4) is followed by calculating the ratio of the long side to the short side of the ellipse of each insulator umbrella skirt, and selecting the flattest ellipse as the ellipse of the target insulator umbrella skirt obtained by final identification.
5. The visual identification method for the insulator cleaning robot of the transformer substation according to any one of claims 1 to 4, characterized by further comprising the step of classifying insulator contamination, wherein the detailed steps comprise:
s1) collecting an insulator image;
s2) filtering the background interference and noise of the insulator image and dividing the image into insulator sub-images;
s3) respectively converting the RGB images of the insulator subgraphs into an LAB color model and an HSI color model, and taking the mean value of the LAB color model and the HSI color model as a color characteristic value;
s4) inputting the color characteristic value into a machine learning classification model which is trained in advance to obtain insulator pollution classification results of each insulator, wherein the machine learning classification model is trained in advance to establish mapping between the color characteristic value of each insulator and the insulator pollution classification results.
6. The visual recognition method for the substation insulator cleaning robot according to claim 5, wherein the machine learning classification model is an SVM classifier, and the step S4) is preceded by a step of performing parameter optimization on the SVM classifier by using a genetic algorithm to determine RBF kernel function c penalty parameters and g kernel function parameters, so that the sample training data which is linearly indivisible in the low-dimensional space is mapped to the high-dimensional feature space through the RBF kernel function c to become linearly separable.
7. A visual identification system for a substation insulator cleaning robot, comprising a computer device, characterized in that the computer device is programmed or configured to perform the steps of the visual identification method for a substation insulator cleaning robot of any one of claims 1 to 6, or that a computer program is stored on a memory of the computer device, which is programmed or configured to perform the visual identification method for a substation insulator cleaning robot of any one of claims 1 to 6.
8. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program programmed or configured to perform the visual identification method for a substation insulator cleaning robot according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911175633.5A CN110909751B (en) | 2019-11-26 | 2019-11-26 | Visual identification method, system and medium for transformer substation insulator cleaning robot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911175633.5A CN110909751B (en) | 2019-11-26 | 2019-11-26 | Visual identification method, system and medium for transformer substation insulator cleaning robot |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110909751A CN110909751A (en) | 2020-03-24 |
| CN110909751B true CN110909751B (en) | 2022-09-02 |
Family
ID=69819650
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911175633.5A Active CN110909751B (en) | 2019-11-26 | 2019-11-26 | Visual identification method, system and medium for transformer substation insulator cleaning robot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110909751B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113009416B (en) * | 2021-04-08 | 2024-03-12 | 国网江苏省电力有限公司检修分公司 | Insulator detection positioning method based on laser sensor array |
| CN114638883B (en) * | 2022-03-09 | 2023-07-14 | 西南交通大学 | Visual limited repositioning target method for insulator water flushing robot |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216895A (en) * | 2007-12-26 | 2008-07-09 | 北京航空航天大学 | An automatic extracting method for ellipse image features in complex background images |
| CN105678760A (en) * | 2016-01-04 | 2016-06-15 | 国家电网公司 | Method for recognizing insulator image on the basis of Canny edge detection algorithm |
| CN107067018A (en) * | 2016-12-09 | 2017-08-18 | 南京理工大学 | A kind of hot line robot bolt recognition methods based on random Hough transformation and SVM |
| CN107240095A (en) * | 2017-05-25 | 2017-10-10 | 武汉大学 | A kind of DC line pollution severity of insulators state recognition method based on visible images |
| CN109785285A (en) * | 2018-12-11 | 2019-05-21 | 西安工程大学 | A kind of insulator damage testing method based on oval feature fitting |
-
2019
- 2019-11-26 CN CN201911175633.5A patent/CN110909751B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216895A (en) * | 2007-12-26 | 2008-07-09 | 北京航空航天大学 | An automatic extracting method for ellipse image features in complex background images |
| CN105678760A (en) * | 2016-01-04 | 2016-06-15 | 国家电网公司 | Method for recognizing insulator image on the basis of Canny edge detection algorithm |
| CN107067018A (en) * | 2016-12-09 | 2017-08-18 | 南京理工大学 | A kind of hot line robot bolt recognition methods based on random Hough transformation and SVM |
| CN107240095A (en) * | 2017-05-25 | 2017-10-10 | 武汉大学 | A kind of DC line pollution severity of insulators state recognition method based on visible images |
| CN109785285A (en) * | 2018-12-11 | 2019-05-21 | 西安工程大学 | A kind of insulator damage testing method based on oval feature fitting |
Non-Patent Citations (2)
| Title |
|---|
| A fast and effective ellipse detector for embedded vision applications;Michele Fornaciari等;《Pattern Recognition》;20140327;第3693-3780页 * |
| 基于红外成像技术的零值绝缘子检测;谭孝元;《中国优秀硕博士学位论文全文数据库（硕士） 工程科技II辑》;20170115;第26-38页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110909751A (en) | 2020-03-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110163853B (en) | Edge defect detection method | |
| CN107909081B (en) | Method for quickly acquiring and quickly calibrating image data set in deep learning | |
| CN108182383B (en) | Vehicle window detection method and device | |
| CN106960178B (en) | Training method of insulator recognition model and insulator recognition and positioning method | |
| CN109409205A (en) | Video road driveway line detecting method of taking photo by plane based on line pitch characteristics point cluster | |
| TWI611961B (en) | Device, method, and computer-readable medium for analyzing lane line image | |
| CN109376740A (en) | A kind of water gauge reading detection method based on video | |
| CN111242899B (en) | Image-based flaw detection method and computer-readable storage medium | |
| CN110909751B (en) | Visual identification method, system and medium for transformer substation insulator cleaning robot | |
| CN111080691A (en) | Infrared hot spot detection method and device for photovoltaic module | |
| CN110866430A (en) | License plate recognition method and device | |
| CN111369516A (en) | Transformer bushing heating defect detection method based on infrared image recognition | |
| CN113240623B (en) | Pavement disease detection method and device | |
| CN110751619A (en) | Insulator defect detection method | |
| CN108508023B (en) | Defect detection system for contact end jacking bolt in railway contact network | |
| CN104915642A (en) | Method and apparatus for measurement of distance to vehicle ahead | |
| CN115841633A (en) | Power tower and power line associated correction power tower and power line detection method | |
| CN109492544B (en) | Method for classifying animal fibers through enhanced optical microscope | |
| CN110427979B (en) | Road water pit identification method based on K-Means clustering algorithm | |
| CN110348307B (en) | Path edge identification method and system for crane metal structure climbing robot | |
| CN111415330A (en) | Copper foil appearance defect detection method based on deep learning | |
| CN111008956A (en) | Beam bottom crack detection method, system, device and medium based on image processing | |
| CN110826571A (en) | Image traversal algorithm for image rapid identification and feature matching | |
| CN116596921B (en) | Method and system for sorting incinerator slag | |
| CN117197700A (en) | Intelligent unmanned inspection contact net defect identification 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 |