A visually assisted inspection system for power cable insulation layer

Abstract

This invention relates to the field of image analysis technology, specifically to a visual-assisted inspection system for power cable insulation layers. The system includes a memory and a processor. The processor executes a computer program stored in the memory to perform the following steps: acquiring the torn area of ​​the insulation layer of the power cable to be inspected; analyzing all edge pixels on the boundary of the torn area to determine the damage diffusion index; filtering out the loop edges within the torn area; determining the damage depth index based on the distance between the loop edges of each cluster and the grayscale changes of the torn area and its centroid; and determining the tear detection result based on the damage diffusion index and the damage depth index. This invention improves the detection accuracy of the visual-assisted inspection system for power cable insulation layers by quantifying the damage diffusion index and the damage depth index, which is beneficial for subsequent appropriate tear repair measures and improves repair efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a visually assisted detection system for an insulation layer of a power cable. Background Art

[0002] Cable is usually a kind of power transmission equipment composed of several or several groups of wires. Its basic structure includes core (conductor), insulation layer, shielding layer and protective layer. The insulation layer is the material wrapped around the conductor to prevent the current from flowing into the external environment. During the production process of power cables, the power cables are affected by many factors such as material quality, production process, operation mode and environmental control, which may cause the cable insulation layer to tear. Once the insulation layer is torn, the power cable will leak current, short circuit and cause fire. Therefore, it is necessary to detect the insulation layer of the power cable to ensure the safety of the power system.

[0003] Different severity levels of tear defects correspond to different treatment methods. When currently detecting tear defects in the insulation layer of a power cable, only defects on the surface can be detected, but the severity of the defects is not clear, resulting in low detection accuracy of the detection system for the insulation layer of the power cable. For tear defects of different severity levels, appropriate repair measures cannot be adopted to achieve repair, affecting the repair efficiency. Summary of the invention

[0004] In order to solve the technical problem that the above-mentioned existing detection system has low detection accuracy for the insulation layer of power cables, the purpose of the present invention is to provide a visual auxiliary detection system for the insulation layer of power cables. The technical solution adopted is as follows:

[0005] An embodiment of the present invention provides a visual aid detection system for an insulation layer of a power cable, comprising a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the following steps:

[0006] Obtain the insulation tearing area of ​​the power cable to be tested;

[0007] Analyze the diffusion of the torn area according to all edge pixels on the boundary of the torn area of ​​the insulation layer, and determine the damage diffusion index of the torn area of ​​the insulation layer;

[0008] Screen out the snare edges in the insulation layer tearing area, perform cluster analysis on the snare edges to obtain clusters; determine the damage depth index of the insulation layer tearing area according to the distance of the snare edges between the clusters, the grayscale change of the insulation layer tearing area and its centroid area;

[0009] The tear detection result of the power cable to be detected is determined according to the damage diffusion index and the damage depth index of the insulation layer tear area.

[0010] Furthermore, the analysis of the diffusion of the torn region according to all edge pixels on the boundary of the torn region of the insulating layer to determine the damage diffusion index of the torn region of the insulating layer includes:

[0011] Determine the damage diffusion factors of the two endpoints according to the position distribution of edge pixel points around the two endpoints on the boundary of the insulation layer tearing area; wherein the two endpoints are the two edge pixel points with the largest distance on the insulation layer tearing boundary;

[0012] Taking the line connecting the two endpoints as a dividing line, dividing the insulation layer tear boundary into two edges by using the dividing line; performing straight line fitting on the two edges to determine fitting loss values, and taking the average of the two fitting loss values ​​as the tear boundary loss value;

[0013] The damage diffusion index of the insulation layer tear area is determined according to the damage diffusion factor, the tear boundary loss value, and the angle between the tear damage direction line and the central axis of the power cable to be tested; wherein the tear damage direction line is the direction of the line connecting the two end points on the insulation layer tear boundary.

[0014] Furthermore, the determining of the damage diffusion factors of the two endpoints according to the position distribution of edge pixel points around the two endpoints on the boundary of the insulation layer tearing region includes:

[0015] For any endpoint, select several edge pixel points on both sides of the endpoint on the tear boundary of the insulation layer, and then determine the fitting straight line corresponding to the edge pixel points selected on each side, and take the angle between the two fitting straight lines intersecting at the endpoint as the endpoint angle;

[0016] Determine each extension angle according to the position distribution relationship between the edge pixels on both sides of the endpoint;

[0017] The damage diffusion factor of the endpoint is determined according to the angle difference between the latter extension angle and the former extension angle and the endpoint angle; wherein the endpoint angle is negatively correlated with the damage diffusion factor.

[0018] Furthermore, the determining of each extension angle according to the position distribution relationship between edge pixels on both sides of the endpoint includes:

[0019] The sliding window slides on the edge pixels on both sides of the endpoint according to the preset step size, determines the window area for each sliding, and marks the corresponding window areas on both sides with serial numbers according to their distance from the endpoint;

[0020] Linear fitting is performed on the edge pixels in each window area to obtain each fitting line, and the angle corresponding to the intersection of two fitting lines with the same window area number is taken as the extension angle.

[0021] Further, screening out each snare edge in the insulation layer tearing area includes:

[0022] Determine each pending edge in the insulation layer tearing region, and map the pending edge to an edge on a boundary of the insulation layer tearing region as a projection edge corresponding to the pending edge;

[0023] The similarity between each pending edge and its projected edge is calculated, and the pending edge with a similarity not less than a preset threshold is taken as a trap edge.

[0024] Furthermore, the clustering analysis of each trap edge to obtain each clustering cluster includes:

[0025] Determine the clustering index of each snare edge according to the distance between each snare edge and its projection edge and the distance between each snare edge and the torn area of ​​the insulating layer;

[0026] According to several clustering K values, all the trap edges are clustered according to the clustering index of each trap edge, and each cluster cluster corresponding to each clustering K value is determined;

[0027] According to the uniform distribution of the edge angles of all the trap edges in each cluster corresponding to each cluster K value, the clustering effect value of each cluster K value is determined; each cluster corresponding to the cluster K value corresponding to the maximum clustering effect value is used as the final cluster;

[0028] The line between the center point of the edge of the snare and the center point of the torn area of ​​the insulating layer is determined as the center line, and the angle between the center line and the horizontal line is recorded as the edge angle.

[0029] Further, the determining of the clustering index of each snare edge according to the distance between each snare edge and its projection edge and the distance between each snare edge and the torn area of ​​the insulating layer includes:

[0030] Determine the center position of each snare edge, the projected edge of each snare edge, and the torn insulating layer area, and calculate the distance between each snare edge and the torn insulating layer area according to the center position of each snare edge and the torn insulating layer area, and record it as the first center distance;

[0031] According to the center positions of each trap edge and its projection edge, the distance between each trap edge and its projection edge is calculated and recorded as the second center distance;

[0032] The clustering index of each trap edge is determined according to the first center distance and the second center distance of each trap edge; wherein the first center distance is positively correlated with the clustering index, and the second center distance is negatively correlated with the clustering index.

[0033] Furthermore, the damage depth index of the torn insulating layer region is determined according to the distance of the trap edge between each cluster, the grayscale change of the torn insulating layer region and its centroid region, including:

[0034] Record any cluster as a target cluster, record any trap edge in the target cluster as a target edge, and determine the trap edge closest to the target edge in other clusters except the target cluster as the associated edge of the target edge;

[0035] According to the center position of each trap edge in each cluster, the center distance between each trap edge and its associated edge in each cluster is determined, which is recorded as the trap distance;

[0036] A preset size area is constructed with the centroid of the insulation layer tearing area as the center as the centroid area, and the damage depth factor is determined according to the grayscale difference between the centroid area and the insulation layer tearing area;

[0037] According to the average snare distance corresponding to each cluster and the damage depth factor, a damage depth index of the insulation layer tearing area is determined; wherein the average snare distance is negatively correlated with the damage depth index.

[0038] Further, determining the damage depth factor according to the grayscale difference between the centroid area and the insulation layer tearing area includes:

[0039] Calculate the grayscale average value corresponding to the centroid area, and calculate the grayscale average value corresponding to the insulation layer tearing area;

[0040] The difference between the grayscale average value corresponding to the centroid area and the grayscale average value corresponding to the torn area of ​​the insulating layer is used as the numerator of the ratio, the grayscale average value corresponding to the torn area of ​​the insulating layer is used as the denominator of the ratio, and the ratio is used as the damage depth factor.

[0041] Further, the determining of the tear detection result of the power cable to be detected according to the damage diffusion index and the damage depth index of the insulation layer tear area includes:

[0042] The product of the damage diffusion index and the damage depth index is calculated, the product is normalized, and the result value after the normalization is used as the tear detection result of the power cable to be detected.

[0043] The present invention has the following beneficial effects:

[0044] The present invention provides a visual aid detection system for the insulation layer of a power cable. The system first obtains the torn area of ​​the insulation layer, and then combines the torn boundary characteristics to quantify the damage diffusion index of the torn area of ​​the insulation layer. The damage diffusion index is determined by multiple calculation factors related to the torn boundary characteristics. Therefore, compared with using a single calculation factor to analyze the damage diffusion situation, the damage diffusion index determined by the present invention considers more comprehensive factors and has higher numerical accuracy. The accurate damage diffusion index can further enhance the reference value of the visual aid detection result in the tear repair process. By performing cluster analysis on the edges of each snare to obtain each cluster cluster, the clusters belonging to the same The edges of the damaged loops are clustered together, which is beneficial to the subsequent calculation of the damage depth index. When determining the damage depth index, not only the position characteristics of the loop edges are combined, but also the grayscale characteristics related to the damage depth are taken into account. While improving the numerical reliability of the damage depth index, it further improves the reference value of the tear detection results in the tear repair process. By determining the tear detection results of the power cable to be inspected, the function of the visual-assisted inspection system is improved, which is also helpful for the subsequent adaptive determination of the repair method of the torn area of ​​the insulation layer based on the detection results, which saves the cost of power cable repair work to a certain extent and improves the efficiency of the repair work. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 This is an execution flow chart of a visual auxiliary detection system for power cable insulation layer according to an embodiment of the present invention;

[0047] Figure 2 Schematic diagram of the process of step S2 in an embodiment of the present invention;

[0048] Figure 3 is a schematic diagram of an end point angle and an extension angle in an embodiment of the present invention;

[0049] Figure 4 is a schematic diagram of a projection edge in an embodiment of the present invention;

[0050] Figure 5 Schematic diagram of the process of step S32 in an embodiment of the present invention;

[0051] Figure 6 Schematic diagram of the process of step S4 in the embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the technical solutions proposed by the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0053] 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.

[0054] The application scenarios targeted by the present invention may be:

[0055] The insulation layer of power cables is the key part of the organization of current leakage. A good insulation layer can prevent current leakage and ensure the safe and stable operation of the power system. Therefore, it is necessary to perform visual inspection on the surface insulation layer of power cables. In order to ensure the quality of the insulation layer of power cables and the safe and stable operation of power systems, machine vision is used to inspect the insulation layer of power cables. However, the existing inspection system is difficult to accurately assess the true impact of tear defects on the overall quality and performance of the insulation layer, resulting in incomplete inspection results of the visual-assisted inspection system, which is not conducive to the subsequent adaptive repair of tear defects.

[0056] In order to determine the severity of each tear area and implement adaptive repair work, this embodiment provides a visually assisted detection system for the insulation layer of a power cable, including a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the following steps:

[0057] Obtain the insulation tearing area of ​​the power cable to be tested;

[0058] Analyze the diffusion of the torn area according to all edge pixels on the boundary of the torn area of ​​the insulation layer, and determine the damage diffusion index of the torn area of ​​the insulation layer;

[0059] Screen out the snare edges in the insulation layer tearing area, perform cluster analysis on the snare edges to obtain clusters; determine the damage depth index of the insulation layer tearing area according to the distance of the snare edges between the clusters, the grayscale change of the insulation layer tearing area and its centroid area;

[0060] The tear detection result of the power cable to be detected is determined according to the damage diffusion index and the damage depth index of the insulation layer tear area.

[0061] The following is a detailed explanation of each of the above steps:

[0062] refer to Figure 1 , shows an execution flow chart of a visual auxiliary detection system for the insulation layer of a power cable according to the present invention, comprising the following steps:

[0063] S1, obtaining the insulation tearing area of ​​the power cable to be detected.

[0064] In order to facilitate the corresponding repair operations on different insulation layer tearing areas, it is necessary to quantify the tearing severity of the insulation layer tearing area. The severity can be analyzed from two aspects, namely, the spread of damage in the tearing area and the deepening of the damage depth. However, it is necessary to obtain the insulation layer tearing area of ​​the power cable to be tested first, because the insulation layer tearing area is the basic data for subsequent data processing and analysis.

[0065] The above step S1 can be implemented by the following steps S11 to S12 (not shown):

[0066] S11, obtaining a grayscale image of the surface of the insulation layer of the power cable to be tested.

[0067] Specifically, first, obtain the power cable to be inspected, which is recorded as the power cable to be inspected; secondly, since the power cable is bent during use, in order to avoid the tearing defect being obscured, the power cable to be inspected is straightened, and the insulation layer on the surface of the power cable to be inspected is photographed using an industrial camera to obtain an insulation layer surface image; then, since the insulation layer tearing defect detection and analysis only involves defect feature detection, there is no need to analyze from the perspective of color space, so the insulation layer surface image must be grayscaled to obtain an insulation layer surface grayscale image.

[0068] Among them, the implementation means of grayscale processing include but are not limited to: weighted average method, average method, maximum method and minimum method, etc. The implementation process of grayscale processing is existing technology and will not be elaborated in detail here.

[0069] S12, performing connected domain analysis on the grayscale image of the surface of the insulation layer to obtain a torn area of ​​the insulation layer.

[0070] In this embodiment, the tear defect usually forms an obvious crack or linear damage, and the boundary between it and the surrounding normal area is often relatively clear. The reason is that the structural change of the torn part leads to the change of the surface reflected light, thus forming an obvious contrast in the image; and the area of ​​the tear defect usually shows a different grayscale value from the surrounding area. The reason is that the light reflection or transmission properties of the torn or cracked part change, resulting in a change in brightness or color in the image.

[0071] Specifically, firstly, combining the edge features and grayscale features of the above-mentioned tearing defects, a connected domain analysis is performed on the grayscale image of the surface of the insulation layer to determine whether there is a tearing defect in the power cable to be tested; secondly, in order to overcome the existing technical problem of unclear severity of the tearing defect, if the power cable to be tested is a power cable with a tearing defect in the insulation layer, first perform the Canny edge detection algorithm on the grayscale image of the surface of the insulation layer, and then combine the characteristics of the lower grayscale value of the tearing area to obtain the tearing connected domain, which is recorded as the tearing area of ​​the insulation layer.

[0072] Among them, the Canny edge detection algorithm and the connected domain acquisition method are both existing technologies, which are not within the scope of protection of the present invention and will not be elaborated in detail here. Of course, the tearing area in the grayscale image of the surface of the insulation layer can also be extracted by other means, which will not be specifically expanded here.

[0073] It should be noted that there may be multiple insulation layer tearing regions, but each insulation layer tearing region has its corresponding tearing detection result, so this embodiment is analyzed by taking the case where there is only one insulation layer tearing region as an example.

[0074] So far, the present embodiment has obtained the insulation layer tearing area of ​​the power cable to be detected.

[0075] S2, analyzing the diffusion of the torn area according to all edge pixels on the boundary of the torn area of ​​the insulation layer, and determining the damage diffusion index of the torn area of ​​the insulation layer.

[0076] First of all, it should be noted that cracks of different severity have different effects on the performance of the insulation layer. The severity of the tear defect can be quantified by combining the image features of the tear defect. Among them, the shape of the damaged part corresponding to the tear defect caused by stress is irregular. When the stress spreads, the damaged part will deepen, that is, the damaged part shows a diffusion trend, which leads to an increase in the severity of the insulation layer tear. Therefore, in order to analyze the severity of the tear defect, it is necessary to predict the probability of the tear area spreading based on the image features generated by the influence of stress on the insulation layer tear, and the damage diffusion index can characterize the possibility of continued cracking and damage in the insulation layer tear area.

[0077] The above step S2 can be Figure 2 The steps shown achieve:

[0078] S21, determining damage diffusion factors of the two endpoints according to the position distribution of edge pixel points around the two endpoints on the boundary of the insulation layer tearing area.

[0079] It should be noted that the tearing stress causes the torn part to present strip-like damage that is wider inside and thinner on both sides. When obtaining the surface image of the insulation layer, the power cable to be inspected was straightened. The straightening operation caused the strip-like damage to squeeze each other, resulting in a reduction in the area of ​​the damaged part. The image characteristics presented in the torn area of ​​the insulation layer are as follows: the angle of the endpoint angle at the endpoint position is small, but the angle of the extension angle at the endpoint position shows a continuous increasing trend.

[0080] The above step S21 can be implemented through steps S211 to S213 (not shown in the figure):

[0081] S211, for any endpoint, select a number of edge pixel points on both sides of the endpoint on the tear boundary of the insulation layer, and then determine the fitting straight lines corresponding to the edge pixel points selected on each side, and take the angle at which the two fitting straight lines intersect at the endpoint as the endpoint angle.

[0082] In this embodiment, the implementation steps of determining the endpoint angles of the two endpoints are consistent. In order to reduce unnecessary descriptions and facilitate understanding of the solution, one of the endpoints is taken as an example to determine the endpoint angle. The implementation steps of determining the endpoint angle may include:

[0083] First, in step S12, all edges corresponding to the insulating layer tearing area are determined. For all the edges, the edge with the largest number of edge pixels is taken as the boundary of the insulating layer tearing area, which can also be called the insulating layer tearing boundary; secondly, for every two edge pixels on the boundary, the Euclidean distance between every two edge pixels is calculated, and the two edge pixels with the largest Euclidean distance are taken as endpoints; then, on the insulating layer tearing boundary, 5 adjacent edge pixels on both sides of the endpoint are selected, that is, the selected edge pixels are the 10 edge pixels closest to the endpoint; then, the least squares method is used to perform straight line fitting on the 5 edge pixels on each side to obtain two fitted straight lines, and the angle formed by the intersection of the two fitted straight lines is taken as the endpoint angle.

[0084] Among them, the implementation process of the least squares method is a prior art and is not within the scope of protection of the present invention, and will not be elaborated on in detail here; regarding the number of edge pixel points selected on one side of the endpoint, the implementer can set it according to the actual situation, and no specific limitation is made here; in addition, since the analysis of the endpoint angle is to analyze the tear expansion at the endpoint position of the tear area of ​​the insulation layer, the endpoint angle here is the minimum angle formed by the intersection of two fitting straight lines. Under normal circumstances, the endpoint angle of the tear area is not greater than 90 degrees.

[0085] S212, based on the position distribution relationship between edge pixels on both sides of the endpoint.

[0086] It should be noted that the endpoint angle of the tear defect is greatly affected by the straightening of the cable wire. The stress causes the two sides of the tear damage site to gradually expand outward. The extension angle of the endpoint shows an increasing trend within a certain range. In order to facilitate the subsequent use of the image features of the extension angle to quantify the damage diffusion factor of the endpoint, it is necessary to determine the various extension angles of the endpoint. The specific implementation process may include:

[0087] First, the sliding window slides on the edge pixels on both sides of the endpoint according to the preset step size, determines the window area for each sliding, and marks the corresponding window areas on both sides with serial numbers according to their distance from the endpoint.

[0088] In this embodiment, the size of the sliding window is empirically set to 5 pixels, and the preset step size of the window sliding is empirically set to 1 pixel. The two edge pixels second closest to the endpoint are taken as the starting points, and the window slides on the edge pixels on both sides of the endpoint. A group of edge pixels can be obtained each time it slides, and 5 groups of edge pixels can be obtained on one side. Each group of edge pixels is marked with a serial number according to the distance from the endpoint.

[0089] For example, for the edge pixels on one side of the endpoint, a group of edge pixels corresponding to the window area closest to the endpoint is marked as the first group, followed by the second group, the third group, the fourth group and the fifth group. When the endpoint angle is determined, the corresponding group of edge pixels on this side is

[01234] , and the first group of edge pixels corresponding to the extension angle of the endpoint is determined as

[12345] , the second group is

[23456] , the third group is

[34567] , the fourth group is

[45678] , and the fifth group is

[56789] .

[0090] Secondly, straight line fitting is performed on the edge pixels in each window area to obtain each fitting line, and the angle corresponding to the intersection of two fitting lines with the same window area number is taken as the extension angle.

[0091] In this embodiment, the least squares method is used to perform straight line fitting on each group of edge pixels to obtain each fitted straight line. For two fitted straight lines with the same window area number, that is, the window area with the same distance relative to the endpoints, the angle formed by the intersection of the two is used as the extension angle of the endpoint.

[0092] The schematic diagram of the end angle and the extension angle is as follows: Figure 3 As shown, in Figure 3 The angle includes an endpoint angle 301 , a first extension angle 302 and a second extension angle 303 .

[0093] S213, determining a damage diffusion factor of the endpoint according to the angle difference between the latter extension angle and the former extension angle and the endpoint angle.

[0094] In this embodiment, the damage diffusion factor of the endpoint is determined by combining the angle difference between two adjacent extension angles and the endpoint angle, wherein the angle difference is positively correlated with the damage diffusion factor, and the larger the angle difference, the larger the damage diffusion factor; the endpoint angle is negatively correlated with the damage diffusion factor, and the smaller the endpoint angle, the more obvious the tearing defect characteristics are, and the larger the damage diffusion factor is.

[0095] As an example, the damage spread factor of the endpoint can be calculated as:

[0096] ; In the formula, KK represents the damage diffusion factor of the endpoint, th represents the hyperbolic tangent function, represents the endpoint angle, i represents the serial number of the extension angle difference, I represents the number of extension angle differences, and each extension angle has a corresponding extension angle difference. represents the i+1th extension angle difference of the endpoint, Represents the i-th extension angle difference of the endpoint, and the extension angle difference is equal to the difference between the subsequent extension angle and the previous extension angle.

[0097] In the calculation formula of the damage diffusion factor, the endpoint angle does not have a zero case; the first extension angle difference of the endpoint can be equal to the difference between the first extension angle and the endpoint angle, but the second extension angle difference is equal to the difference between the second extension angle and the first extension angle; when the endpoint angle is smaller and the difference between its extension angles is larger, it indicates that the damaged part is narrower at the endpoint and extends wider and wider outward, and the damage diffusion factor of the endpoint is larger; the hyperbolic tangent function can map the input data to -1 to 1, but due to It is a non-negative number, so the case where the damage diffusion factor is negative is not considered.

[0098] S22, taking the line connecting the two end points as the dividing line, and using the dividing line to divide the tear boundary of the insulation layer into two edges; performing straight line fitting on the two edges to determine the fitting loss value, and taking the average of the two fitting loss values ​​as the tear boundary loss value.

[0099] In this embodiment, the method for determining the fitting loss value is a prior art and is not within the protection scope of the present invention, and will not be elaborated on in detail here.

[0100] S23, determining a damage diffusion index of the torn area of ​​the insulation layer according to the damage diffusion factor, the tear boundary loss value, and the angle between the tear damage direction line and the central axis of the power cable to be tested.

[0101] In this embodiment, the tearing damage direction line is the direction of the line connecting the two end points on the tear boundary of the insulation layer. Combining the three aspects of damage diffusion influencing factors, the damage diffusion index can be quantified. The numerical accuracy of the damage diffusion index determined by multiple factors is higher, which helps to improve the reliability of the detection of the degree of tearing of the cable insulation layer.

[0102] As an example, the calculation formula for the damage diffusion index of the insulation layer tear area can be:

[0103] ; In the formula, SL represents the damage diffusion index of the insulation layer tearing area, th represents the hyperbolic tangent function, Indicates the angle between the tearing direction line and the central axis of the power cable to be tested. represents the tearing boundary loss value, represents the damage spread factor of the jth endpoint.

[0104] In the calculation formula of the damage diffusion index, the unit influence of each calculation factor is not considered, and only the corresponding numerical value is taken to determine the damage diffusion index; in order to facilitate the analysis of the verticality between the insulation layer tearing area and the power cable, The value range of is between 0 and 90 degrees, that is, when determining the angle, the minimum angle is used as the final angle. For example, if there are two angles between the direction line and the central axis, such as 40 and 140 degrees, 40 degrees is selected as the final angle. The larger the value, the more likely it is that the insulation layer tear area is perpendicular to the power cable. Since the possibility of tearing and propagation of vertical tear defects in the power cable is greater during use, it is positively correlated with the damage diffusion index. The parameter 1 is added to avoid extreme cases that cause the damage diffusion index to be zero. The implementer can reset the parameter according to specific needs. The larger it is, the more likely it is that the insulation tearing area is at the bending part of the cable. Since the bending of the cable makes the damaged part more likely to tear, it is positively correlated with the damage diffusion index. The larger the value, the narrower the end point of the damaged part is and the wider it is towards the outside. The stress causes the end point to be more likely to tear, so it is also positively correlated with the damage diffusion index. Of course, the implementer can also use other calculation methods to combine various calculation factors related to the damage diffusion index, for example, by adding together various calculation factors.

[0105] So far, this embodiment has determined the damage diffusion index of the torn area of ​​the insulation layer.

[0106] S3, screening out each snare edge in the insulation layer tearing area, and performing cluster analysis on each snare edge to obtain each cluster.

[0107] First of all, it should be noted that the power cable is made of multiple layers of materials, so there may be multiple layers of damage on the surface of the power cable insulation layer. Therefore, it is necessary to analyze the internal edge of the insulation layer tear area to obtain the damage depth of the tear area. The greater the damage depth, the higher the severity of the tear defect. Among them, the multi-layer damage leads to the formation of a loop between the damaged layers, that is, there are edges in the insulation layer tear area that are similar to the insulation layer tear boundary shape and are nested with each other. The screened loop edges can be used to analyze the damage depth of the tear defect.

[0108] The above step S3 can be implemented through steps S31 to S32 (not shown in the figure):

[0109] S31, screening out the respective snare edges in the insulation layer tearing region.

[0110] First, each pending edge in the insulation layer tearing region is determined, and the pending edge is mapped to an edge on the insulation layer tearing boundary as a projection edge corresponding to the pending edge.

[0111] In this embodiment, the edge detection algorithm can be used to determine each edge in the insulation layer tearing area and define it as a pending edge; determine the center of the insulation layer tearing area, and for any pending edge, connect the pending edge with the center of the insulation layer tearing area to the mapping edge of the insulation layer tearing boundary as the projection edge of the corresponding pending edge, that is, map the pending edge to the insulation layer tearing boundary. Each pending edge has its corresponding projection edge; the schematic diagram of the projection edge is shown in FIG. Figure 4 As shown, Figure 4 The edge detection algorithm includes a pending edge 401 and a projection edge 402 of the pending edge. The implementation process of the edge detection algorithm is a prior art and is not within the protection scope of the present invention, and will not be elaborated here.

[0112] Secondly, the similarity between each pending edge and its projected edge is calculated, and the pending edge with a similarity not less than a preset threshold is taken as a trap edge.

[0113] In this embodiment, the edge features of the trap edge and the edge features of the insulation layer tear boundary have similar characteristics, so the DTW (Dynamic Time Warping) distance algorithm is used to measure the similarity between each pending edge and its projected edge. The larger the DTW value, the worse the similarity between the pending edge and the projected edge. The exp (-) function is used to perform negative correlation normalization processing on the DTW value, and the result value after negative correlation normalization processing is used as the similarity. Among them, the preset threshold can be set to 0.5, and the implementer can set it according to the specific actual situation without specific limitation; the calculation process of the DTW distance is a prior art, which is not within the protection scope of the present invention and will not be elaborated in detail here.

[0114] S32, performing cluster analysis on the edges of each trap to obtain each cluster.

[0115] It should be noted that a single trap edge may not be able to represent the damage of a layer, that is, the damage of the same layer may be affected by multiple trap edges, so it is necessary to perform cluster analysis on all the trap edges and cluster the trap edges with the same layer of damage characteristics into the same cluster. Since different clustering K values ​​correspond to different clustering results, and the clustering results correspond to different clustering effects, in order to improve the numerical reference value of the damage depth index determined later, it is necessary to select several different clustering K values, analyze the clustering situation of each clustering K value, and finally obtain the target clustering K value of the optimal clustering situation and its corresponding clustering cluster.

[0116] The above step S32 can be performed by Figure 5 The steps shown achieve:

[0117] S321, determining a clustering index of each snare edge according to the distance between each snare edge and its projection edge and the distance between each snare edge and the torn region of the insulating layer.

[0118] Specifically, the center positions of each trap edge, the projected edge of each trap edge, and the torn insulating layer area are determined. According to the center positions of each trap edge and the torn insulating layer area, the distance between each trap edge and the torn insulating layer area is calculated using the Euclidean distance algorithm, which is recorded as the first center distance; according to the center positions of each trap edge and its projected edge, the distance between each trap edge and its projected edge is calculated again using the Euclidean distance algorithm, which is recorded as the second center distance; according to the first center distance and the second center distance of each trap edge, the clustering index of each trap edge is determined, the first center distance is positively correlated with the clustering index, and the second center distance is negatively correlated with the clustering index, such as using the ratio of the first center distance to the second center distance as the clustering index. Among them, each trap edge has its corresponding first center distance and second center distance, and in this embodiment, the second center distance does not have a zero case.

[0119] It should be noted that the first center distance can characterize the distance between the edge of the snare and the torn area of ​​the insulation layer, while the second center distance can characterize the distance between the edge of the snare and the torn boundary of the insulation layer. When the first center distance and the second center distance of the two snare edges are close, it means that the two snare edges are likely to belong to the same damaged layer, so the ratio of the first center distance to the second center distance is used as the clustering indicator of the snare edge.

[0120] S322: Clustering all the trap edges according to the clustering index of each trap edge according to a plurality of clustering K values, and determining each cluster cluster corresponding to each clustering K value.

[0121] Specifically, the ratio of the first center distance to the second center distance of the trap edge is calculated to obtain the clustering index, and the clustering index is used as the clustering object. The K-means clustering algorithm is used to perform multiple clustering processes on all the trap edges, and each cluster cluster corresponding to each cluster K value can be obtained. Among them, the implementation process of the K-means clustering algorithm is a prior art and is not within the scope of protection of the present invention, and will not be elaborated in detail here; the cluster K value can be set to 3, 4 and 5, and the implementer can set the size and number of the cluster K value according to the specific actual situation, without specific limitation.

[0122] S323, determining a clustering effect value of each clustering K value according to the uniform distribution of edge angles of all trap edges in each clustering cluster corresponding to each clustering K value.

[0123] Specifically, the information entropy of the edge angles of all the trap edges in each cluster corresponding to each cluster K value is calculated. The information entropy can characterize the uniformity of the distribution of the edge angles, and the information entropy is determined as the clustering effect value corresponding to the cluster K value. The calculation process of the edge angle can be: determine the line between the center point of the trap edge and the center point of the insulation layer tearing area, record it as the center line, and record the angle between the center line and the horizontal line as the edge angle.

[0124] It is worth noting that the edge angle can characterize the positional relationship between the trap edge and the center point of the insulation layer tearing area. The more types of edge angles there are, the more likely the trap edge in the same cluster is to be located in various directions of the center point of the insulation layer tearing area, and the better the clustering effect of the clustering K value will be.

[0125] In this embodiment, for the information entropy of the edge angles of all the trap edges in the cluster, the larger the information entropy, the more complex the edge angles contained in the corresponding cluster, which means that there are more types of edge angles in the corresponding cluster, and the more types can further indicate that the possibility of the trap edges in the corresponding cluster being arranged in a circle is greater, and the possibility that the trap edges in the corresponding cluster belong to the edges of the same layer of damaged materials is more likely, and the clustering effect is better. Among them, the calculation process of information entropy is a prior art, which is not within the protection scope of the present invention, and will not be elaborated here.

[0126] S324: taking the clustering K value corresponding to the maximum clustering effect value as the target clustering K value, and obtaining each cluster cluster corresponding to the target clustering K value.

[0127] In this embodiment, the larger the clustering effect value is, the greater the possibility that each clustering cluster of the clustering K value corresponding to the clustering effect value is an internal damaged edge of a layer. In order to facilitate the subsequent quantification of the damage depth index based on the image features of the internal damaged edges of each layer, it is necessary to use the clustering K value corresponding to the maximum clustering effect value as the target clustering K value, and obtain each clustering cluster corresponding to the target clustering K value.

[0128] It should be noted that the purpose of determining the target clustering K value is to screen out clustering clusters with higher clustering effect. Compared with the clustering clusters determined by randomly setting the clustering K value, the clustering clusters corresponding to the target clustering K value determined in this embodiment are more in line with the damage nesting characteristics of the torn insulation layer, and are more conducive to enhancing the numerical accuracy of the damage depth index calculated subsequently. Therefore, the final clustering result of each clustering cluster corresponding to the target clustering K value is taken.

[0129] So far, this embodiment has obtained each trap edge in each cluster.

[0130] S4, determining the damage depth index of the insulation layer tearing area according to the distance of the trap edge between each cluster, the grayscale change of the insulation layer tearing area and its centroid area.

[0131] First of all, it should be noted that the tearing defects with the same number of damaged layers still have differences in the speed of damage. The faster the damage downwards, the more severe the tearing defect. In order to comprehensively analyze the severity of the damage in the torn area of ​​the insulation layer, it is necessary to quantify the damage depth. When the damaged layers inside the damage are close, it means that the faster the damage stress spreads downward, the greater the damage depth. When there is a large difference between the grayscale value of the centroid of the torn area of ​​the insulation layer and the overall damaged area, it means that the guided exposure range of the power cable is large and the damage depth will also be large.

[0132] The above step S4 can be performed by Figure 6 The steps shown achieve:

[0133] S41, any cluster is recorded as a target cluster, any trap edge in the target cluster is recorded as a target edge, and the trap edge closest to the target edge in other clusters except the target cluster is determined as an associated edge of the target edge.

[0134] In this embodiment, in order to facilitate the analysis of the distance relationship between the trap edge corresponding to each damaged layer and the trap edges corresponding to other damaged layers except itself, any trap edge in any cluster is taken as an example to determine the trap distance of the trap edge. When determining the associated edge, first calculate the distance between the center of the target edge and the center of each trap edge in other clusters except the target cluster, and then take the trap edge with the minimum distance as the associated edge of the target edge; then, refer to the determination method of the associated edge of the target edge, and obtain the associated edge of each trap edge in each cluster. Among them, the calculation method of the distance between two points is a prior art, which is not within the protection scope of the present invention and will not be elaborated here.

[0135] S42, according to the center position of each trap edge in each cluster, determine the center distance between each trap edge and its associated edge in each cluster, and record it as the trap distance.

[0136] In this embodiment, the center distance is the distance between the center point of the trap edge and the center point of its associated edge, which can be recorded as the trap distance. Each trap edge has its corresponding trap distance, and the trap distance can be used to characterize the distance between damaged layers.

[0137] S43, constructing a preset size area as a centroid area with the centroid of the insulation layer tearing area as the center, and determining a damage depth factor according to a grayscale difference between the centroid area and the insulation layer tearing area.

[0138] In this embodiment, the damage depth factor can characterize the wire exposure state in the insulation layer tear area. The larger the damage depth factor, the more wire is exposed in the insulation layer tear area and the greater the damage severity. Combined with the feature that there is a large difference between the grayscale of the exposed wire and the overall grayscale of the damaged part, the damage depth factor can be quantified.

[0139] Specifically, the grayscale average value corresponding to the centroid area is calculated, and the grayscale average value corresponding to the torn insulation layer area is calculated; the difference between the grayscale average value corresponding to the centroid area and the grayscale average value corresponding to the torn insulation layer area is used as the numerator of the ratio, the grayscale average value corresponding to the torn insulation layer area is used as the denominator of the ratio, and the ratio is used as the damage depth factor.

[0140] Among them, the size of the centroid area can be 3*3, and the implementer can set the size of the centroid area according to the actual situation without making specific restrictions; there is no possibility that the grayscale average value corresponding to the torn area of ​​the insulating layer is zero, so the grayscale average value corresponding to the torn area of ​​the insulating layer can be used as the denominator of the ratio.

[0141] S44, determining a damage depth index of the insulation layer tearing area according to the average snare distance and the damage depth factor corresponding to each cluster.

[0142] In this embodiment, the damage depth index of the insulation layer tearing area can be determined by combining two calculation factors related to the damage depth, namely, the average snare distance and the damage depth factor.

[0143] As an example, the calculation formula for the damage depth index of the insulation layer tear area can be:

[0144] ; In the formula, PS represents the damage depth index of the insulation layer tearing area, exp represents the exponential function with the natural constant e as the base, B represents the number of clusters, b represents the cluster number, represents the average trap distance corresponding to the b-th cluster, Indicates the grayscale average value corresponding to the torn area of ​​the insulation layer, Represents the grayscale average value corresponding to the centroid area, represents the damage depth factor, It represents the inverse of the damage depth factor. Since the damage depth factor can be zero, 1 is added to the denominator. Represents the absolute value function.

[0145] In the calculation formula of the damage depth index, The smaller it is, the closer the distance between the damaged traps of each cluster and the damaged traps around it is, which further shows that the closer the distance between the trap-shaped damage inside the damage is, the faster the stress spreads downward, and the larger the damage depth index is. Therefore, the average trap distance is negatively correlated with the damage depth index. The larger the value is, the greater the difference between the grayscale value of the centroid and the grayscale value of the damaged part as a whole. This may be caused by two situations. First, the damage is too deep, resulting in a too deep grayscale value inside the damage, which is too low and has a large difference with the overall grayscale. Second, the damage causes the wire to be exposed, resulting in a higher grayscale value inside the damage, which is significantly different from the overall grayscale value of the damaged part. Therefore, the damage depth factor is positively correlated with the damage depth index. exp (-) is used to realize the normalization processing of the negative correlation of the data, so the value range of the damage depth index is between 0 and 1.

[0146] So far, this embodiment has obtained the damage depth index of the torn area of ​​the insulation layer.

[0147] S5, determining a tear detection result of the power cable to be detected according to a damage diffusion index and a damage depth index of the torn area of ​​the insulation layer.

[0148] In this embodiment, after obtaining the damage diffusion index and the damage depth index of the insulation layer tear area, the two indexes need to be combined, and the combined data is used as the final tear detection result of the power cable to be detected.

[0149] Specifically, the product of the damage diffusion index and the damage depth index is calculated, the product is normalized, and the result value after normalization is used as the tear detection result of the power cable to be detected. Among them, the result value after normalization is the damage severity of the torn area of ​​the insulation layer, and the damage severity is used as one of the tear detection results. The normalization method can be a linear normalization function. The implementation process of the normalization method is a prior art and is not within the protection scope of the present invention, and will not be elaborated here.

[0150] It should be noted that if there are multiple insulation layer tearing areas in the power cable to be tested, the damage severity of each insulation layer tearing area is determined by referring to the calculation process of the damage diffusion index and the damage depth index of the above-mentioned insulation layer tearing area, and the damage severity of all insulation layer tearing areas is used as the tearing detection result of the power cable to be tested.

[0151] It should be further explained that after obtaining the final tear detection result of the power cable to be detected, a suitable repair method can be adaptively selected for different insulation layer tears, specifically including:

[0152] After obtaining the location of the insulation layer tear area and its damage severity, the first damage threshold is set to 0.5. For the insulation layer tear area with a damage severity less than the first damage threshold, a suitable adhesive can be used to repair the damage, and it is ensured that the adhesive is suitable for the material and stress environment; the second damage threshold is set to 0.8. For the insulation layer tear area with a damage severity not less than the first damage threshold but less than the second damage threshold, it is possible to consider using a patch or applying appropriate touch-up paint to repair the surface, and it is necessary to ensure that the repair material is compatible with the cable material and can withstand the normal stress and environmental conditions of the power cable; for the insulation layer tear area with a damage severity not less than the second damage threshold, a new power cable insulation layer can be replaced. Among them, for the first damage threshold and the second damage threshold, the implementer can set the size of the two thresholds according to the actual situation, without specific restrictions.

[0153] So far, this embodiment has achieved the quantification of the tear severity of the power cable to be detected, and improved the detection accuracy of the auxiliary detection system of the power cable insulation layer. It is conducive to the use of adaptive repair methods for different tear areas of the power cable, thereby improving the efficiency of repairing tear defects in the power cable.

[0154] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A visually assisted inspection system for the insulation layer of power cables. It is characterized in that The invention comprises a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the following steps: Obtain the insulation tearing area of ​​the power cable to be tested; Analyze the diffusion of the torn area according to all edge pixels on the boundary of the torn area of ​​the insulation layer, and determine the damage diffusion index of the torn area of ​​the insulation layer; Screen out the snare edges in the insulation layer tearing area, perform cluster analysis on the snare edges to obtain clusters; determine the damage depth index of the insulation layer tearing area according to the distance of the snare edges between the clusters, the grayscale change of the insulation layer tearing area and its centroid area; Determining a tear detection result of the power cable to be detected according to the damage diffusion index and the damage depth index of the insulation layer tear area; The method of analyzing the diffusion of the torn region according to all edge pixels on the boundary of the torn region of the insulating layer to determine the damage diffusion index of the torn region of the insulating layer includes: Determine the damage diffusion factors of the two endpoints according to the position distribution of edge pixel points around the two endpoints on the boundary of the insulation layer tearing area; wherein the two endpoints are the two edge pixel points with the largest distance on the insulation layer tearing boundary; Taking the line connecting the two endpoints as a dividing line, dividing the insulation layer tear boundary into two edges by using the dividing line; performing straight line fitting on the two edges to determine fitting loss values, and taking the average of the two fitting loss values ​​as the tear boundary loss value; According to the damage diffusion factor, the tear boundary loss value and the angle between the tear damage direction line and the central axis of the power cable to be tested, the damage diffusion index of the insulation layer tear area is determined; wherein the tear damage direction line is the direction of the line connecting the two end points on the insulation layer tear boundary; The determining of the damage diffusion factors of the two endpoints according to the position distribution of edge pixel points around the two endpoints on the boundary of the insulation layer tearing region includes: For any endpoint, select several edge pixel points on both sides of the endpoint on the tear boundary of the insulation layer, and then determine the fitting straight line corresponding to the edge pixel points selected on each side, and take the angle between the two fitting straight lines intersecting at the endpoint as the endpoint angle; Determine each extension angle according to the position distribution relationship between the edge pixels on both sides of the endpoint; Determine the damage diffusion factor of the endpoint according to the angle difference between the latter extension angle and the former extension angle and the endpoint angle; wherein the endpoint angle is negatively correlated with the damage diffusion factor; Determining the damage depth index of the torn insulation layer region according to the distance of the trap edge between each cluster, the grayscale change of the torn insulation layer region and its centroid region includes: Record any cluster as a target cluster, record any trap edge in the target cluster as a target edge, and determine the trap edge closest to the target edge in other clusters except the target cluster as the associated edge of the target edge; According to the center position of each trap edge in each cluster, the center distance between each trap edge and its associated edge in each cluster is determined, which is recorded as the trap distance; A preset size area is constructed with the centroid of the insulation layer tearing area as the center as the centroid area, and the damage depth factor is determined according to the grayscale difference between the centroid area and the insulation layer tearing area; Determine the damage depth index of the insulation layer tearing area according to the average snare distance corresponding to each cluster and the damage depth factor; wherein the average snare distance is negatively correlated with the damage depth index; The method of determining the damage depth factor according to the grayscale difference between the centroid area and the insulation layer tearing area includes: Calculate the grayscale average value corresponding to the centroid area, and calculate the grayscale average value corresponding to the insulation layer tearing area; The difference between the grayscale average value corresponding to the centroid area and the grayscale average value corresponding to the torn area of ​​the insulating layer is used as the numerator of the ratio, the grayscale average value corresponding to the torn area of ​​the insulating layer is used as the denominator of the ratio, and the ratio is used as the damage depth factor.

2. A visually assisted detection system for the insulation layer of a power cable according to claim 1, It is characterized in that The step of determining each extension angle according to the position distribution relationship between edge pixels on both sides of the endpoint includes: The sliding window slides on the edge pixels on both sides of the endpoint according to the preset step size, determines the window area for each sliding, and marks the corresponding window areas on both sides with serial numbers according to their distance from the endpoint; Linear fitting is performed on the edge pixels in each window area to obtain each fitting line, and the angle corresponding to the intersection of two fitting lines with the same window area number is taken as the extension angle.

3. A visually assisted detection system for the insulation layer of a power cable according to claim 1, It is characterized in that The step of screening out the respective snare edges in the insulation layer tearing region comprises: Determine each pending edge in the insulation layer tearing region, and map the pending edge to an edge on a boundary of the insulation layer tearing region as a projection edge corresponding to the pending edge; The similarity between each pending edge and its projected edge is calculated, and the pending edge with a similarity not less than a preset threshold is taken as a trap edge.

4. A visually assisted detection system for the insulation layer of a power cable according to claim 3, It is characterized in that The clustering analysis of each trap edge to obtain each cluster cluster includes: Determine the clustering index of each snare edge according to the distance between each snare edge and its projection edge and the distance between each snare edge and the torn area of ​​the insulating layer; According to several clustering K values, all the trap edges are clustered according to the clustering index of each trap edge, and each cluster cluster corresponding to each clustering K value is determined; According to the uniform distribution of the edge angles of all the trap edges in each cluster corresponding to each cluster K value, the clustering effect value of each cluster K value is determined; each cluster corresponding to the cluster K value corresponding to the maximum clustering effect value is used as the final cluster; The line between the center point of the edge of the snare and the center point of the torn area of ​​the insulating layer is determined as the center line, and the angle between the center line and the horizontal line is recorded as the edge angle.

5. A visually assisted detection system for the insulation layer of a power cable according to claim 4, It is characterized in that Determining the clustering index of each snare edge according to the distance between each snare edge and its projection edge and the distance between each snare edge and the torn area of ​​the insulating layer includes: Determine the center position of each snare edge, the projected edge of each snare edge, and the torn insulating layer area, and calculate the distance between each snare edge and the torn insulating layer area according to the center position of each snare edge and the torn insulating layer area, and record it as the first center distance; According to the center positions of each trap edge and its projection edge, the distance between each trap edge and its projection edge is calculated and recorded as the second center distance; The clustering index of each trap edge is determined according to the first center distance and the second center distance of each trap edge; wherein the first center distance is positively correlated with the clustering index, and the second center distance is negatively correlated with the clustering index.

6. A visually assisted detection system for the insulation layer of a power cable according to claim 1, It is characterized in that The step of determining the tear detection result of the power cable to be detected according to the damage diffusion index and the damage depth index of the insulation layer tear area includes: The product of the damage diffusion index and the damage depth index is calculated, the product is normalized, and the result value after the normalization is used as the tear detection result of the power cable to be detected.

Images