A model fusion-based insulator discharge defect adaptive recognition method and system

By combining single-stage and two-stage target detection models and using an adaptive method based on the maximum circumcircle of polygons and rectangles, the problem of drones and inspection robots being unable to automatically identify insulator discharge defects was solved, achieving efficient and accurate identification and risk level determination of insulator discharge defects.

CN116468688BActive Publication Date: 2025-12-05WUHAN SANJIANG CLP TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310400135.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-04-12
Publication Date
2025-12-05
Estimated Expiration
2043-04-12

AI Technical Summary

Technical Problem

In existing technologies, drones and inspection robots cannot automatically identify insulator discharge defects, and the collected videos or images are at an angle, resulting in insufficient identification capabilities and difficulty in accurately determining the severity of discharge defects.

Method used

An adaptive method based on the maximum circumcircle and maximum circumcircle rectangle of polygons is adopted, combined with single-stage and two-stage target detection models, and insulator discharge defects are identified and risk levels are determined through image processing and coordinate data fusion.

Benefits of technology

It enables efficient identification of vertical and tilted insulators, improves detection accuracy, reduces labor costs, and provides adaptive identification and diagnostic support for insulator discharge defects.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116468688B_ABST
    Figure CN116468688B_ABST
Patent Text Reader

Abstract

The application discloses a model fusion-based insulator discharge defect adaptive identification method and system, which is applied to the technical field of online monitoring, obtains a data set through frame-by-frame decomposition of a video, inputs a picture to be inferred for inference after training of a target detection model, and obtains stable insulator coordinates in vertical, horizontal and inclined directions by combining inference results; obtains an ultraviolet image discharge area through color separation and an external polygon; obtains the length of the discharge area along the insulator direction by using an adaptive method based on the insulator inclination angle and the discharge area coordinates; and further determines the insulator discharge defect risk level; the adaptive method based on the polygon maximum circumscribed circle and the maximum circumscribed rectangle and the joint solution method can effectively identify the discharge defects of the insulators in the video and the picture and correctly determine the insulator discharge defect risk level, thereby providing important support basis for further insulator discharge defect diagnosis.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The present application relates to the online monitoring technical field, more particularly, to a model fusion-based insulator discharge defect adaptive identification method and system. BACKGROUND

[0002] At present, the insulator is a special insulating control installed between conductors at different potentials or conductors and ground components to achieve electrical insulation and mechanical fixation. The internal defects of the insulator have different ultraviolet performance characteristics at different development stages. For example, when the internal dry rot defect develops to the point where the sheath is damaged, obvious ultraviolet discharge will occur. When serious dirt accumulation causes creepage, the insulator surface will also have serious ultraviolet discharge, and even cause insulation breakdown, thereby damaging the use of the entire line.

[0003] Therefore, real-time monitoring of the insulator discharge process, judging the size of the discharge area, and diagnosing the discharge defect level through the discharge length are crucial for the stable operation of long-distance power transmission on overhead lines. However, the current defect inspection of insulators mainly uses unmanned aerial vehicles for power transmission lines, and intelligent inspection robots for substations. However, both methods rely on manual judgment of whether the insulator has discharge defects and the severity of the discharge defects after collecting a large number of videos or pictures on site. Neither the cruise unmanned aerial vehicle nor the inspection robot has the ability to automatically identify discharge defects and their severity. Moreover, the collected insulator videos or pictures have various tilt angles, which require high recognition capabilities of the machine. SUMMARY

[0004] To address at least one of the deficiencies in the prior art or the need for improvement, the present application provides a model fusion-based insulator discharge defect adaptive identification method and system, which uses a combination of adaptive methods based on the maximum circumscribed circle of a polygon and the maximum circumscribed rectangle to efficiently identify discharge defects in videos and pictures and correctly determine the risk level of insulator discharge defects, providing important support for further insulator discharge defect diagnosis.

[0005] To achieve the above-mentioned purpose, according to the first aspect of the present application, a model fusion-based insulator discharge defect adaptive identification method is provided, which comprises:

[0006] Obtaining an insulator picture dataset, performing first image processing on the picture dataset to obtain a first sample set, and constructing a first target detection model based on the first sample set, wherein the first target detection model is a single-stage target detection model;

[0007] performing second image processing on the picture data set to obtain a second sample set, and constructing a second target detection model according to the second sample set, the second target detection model being a two-stage target detection model;

[0008] inputting picture data of the insulator to be detected under the same acquisition means as the picture data set of the insulator to the first target detection model to obtain first coordinate data, and to the second target detection model to obtain second coordinate data, and analyzing the second coordinate data to determine state information of the insulator to be detected;

[0009] performing fusion analysis on the first coordinate data and the second coordinate data to obtain fusion coordinate data of the insulator to be detected, and calculating length data of the insulator to be detected according to the fusion coordinate data;

[0010] determining an external polygon contour of a discharge region from ultraviolet picture data in the picture data of the insulator to be detected, and calculating a first length of the discharge region of the insulator to be detected according to the external polygon contour and the fusion coordinate data;

[0011] selecting a first length in the external polygon contour that is the same as the direction of the insulator to be detected, and calculating a length proportion of the discharge region on the surface of the insulator to be detected according to the first length and the length data.

[0012] Further, the step of determining the state information of the insulator to be detected according to the second coordinate data comprises:

[0013] analyzing the second coordinate data, calculating the inclination angle of the insulator to be detected from the second coordinate data, and classifying the state to obtain a state classification result, the state classification result including a first state and a second state of the insulator to be detected;

[0014] the first state being a vertical or horizontal state under the same marking means as the second coordinate data;

[0015] the second state being an inclined state under the same marking means as the second coordinate data.

[0016] Further, the step of performing fusion analysis on the first coordinate data and the second coordinate data to obtain the fusion coordinate data of the insulator to be detected comprises:

[0017] if the state classification result of the insulator to be detected is the first state, the first coordinate data is taken as the fusion coordinate data of the insulator to be detected;

[0018] If the state classification result of the to-be-detected insulator is the second state, the first coordinate data and the second coordinate data are subjected to intersection-over-union calculation, and the second coordinate data with the maximum intersection-over-union is obtained as the fusion coordinate data.

[0019] Further, the step of determining the inscribed polygon contour of the discharge region in the ultraviolet picture data in the picture data of the to-be-detected insulator comprises:

[0020] The discharge region in the ultraviolet picture data in the picture data of the to-be-detected insulator is determined, a polygon approximation method is used to perform contour surrounding on the discharge region, and an inscribed polygon contour of the discharge region is obtained; the inscribed polygon contour comprises a maximum inscribed rectangle and a maximum inscribed circle.

[0021] Further, the step of calculating the first length of the discharge region of the to-be-detected insulator according to the inscribed polygon contour and the fusion coordinate data comprises:

[0022] If the state classification result of the to-be-detected insulator is the first state, the difference between the maximum and minimum coordinates of the inscribed polygon contour on the first direction axis or the second direction axis is calculated as the first length.

[0023] Further, the step of calculating the first length of the discharge region of the to-be-detected insulator according to the inscribed polygon contour and the fusion coordinate data further comprises:

[0024] If the state classification result of the to-be-detected insulator is the second state, the first maximum length along the first direction axis and the second maximum length along the second direction axis of the discharge region are determined, the first direction axis and the second direction axis are perpendicular; the first maximum length is the difference between the maximum and minimum coordinates of the inscribed polygon contour on the first direction axis; and the second maximum length is the difference between the maximum and minimum coordinates of the inscribed polygon contour on the second direction axis.

[0025] When the first maximum length is greater than or equal to the second maximum length, it is determined that the polygon contour is a maximum inscribed rectangle, and the length of the cutting line parallel to the insulator is calculated as the first length.

[0026] When the second maximum length is greater than or equal to the first maximum length, it is determined that the polygon contour is a maximum inscribed circle, and the diameter of the maximum inscribed circle is calculated as the first length.

[0027] Further, the calculation formula of the length ratio is as follows:

[0028]

[0029] wherein, L jyz is the length of the target insulator, L contours is the first length of the discharge region of the insulator to be detected.

[0030] According to a second aspect of the present application, there is also provided an insulator discharge defect adaptive recognition system based on model fusion, characterized in that it comprises:

[0031] a first image processing unit for obtaining a picture data set of an insulator, performing first image processing on the picture data set to obtain a first sample set, and constructing a first target detection model based on the first sample set, the first target detection model being a single-stage target detection model;

[0032] a second image processing unit for performing second image processing on the picture data set to obtain a second sample set, and constructing a second target detection model based on the second sample set, the second target detection model being a two-stage target detection model;

[0033] a picture processing unit for inputting picture data of an insulator to be detected under the same acquisition means as the picture data set of the insulator to the first target detection model to obtain first coordinate data, and to the second target detection model to obtain second coordinate data, and analyzing the second coordinate data to determine state information of the insulator to be detected;

[0034] a fusion data analysis unit for performing fusion analysis on the first coordinate data and the second coordinate data to obtain fusion coordinate data of the insulator to be detected, and calculating the length data of the insulator to be detected based on the fusion coordinate data;

[0035] a discharge region analysis unit for determining the outer polygon contour of the discharge region from ultraviolet picture data in the picture data of the insulator to be detected, and calculating the first length of the discharge region of the insulator to be detected based on the outer polygon contour and the fusion coordinate data;

[0036] an evaluation unit for selecting the first length of the outer polygon contour that is in the same direction as the insulator to be detected, and calculating the length proportion of the discharge region on the surface of the insulator to be detected based on the first length and the length data.

[0037] Overall, the above technical solutions conceived by the present application can achieve the following beneficial effects compared with the prior art:

[0038] (1) The application can stably detect vertical and horizontal insulators, and can detect the coordinates and inclination angle of the inclined insulator in real time, so as to calculate the length proportion of the discharge area along the inclination angle of the insulator, and then effectively determine the insulator discharge area, accurately calculate the length proportion of the discharge area of the ultraviolet image, correctly determine the risk level of the insulator discharge defect, and realize the adaptive recognition of the insulator discharge defect. It provides an important support basis for further insulator discharge defect diagnosis.

[0039] (2) The application can intelligently recognize the insulator information of the overhead line through the training of at least two deep learning target detection models, improve the efficiency and accuracy of detecting vertical and horizontal insulators while ensuring high stability, ensure the data integrity and accuracy of the inclined insulator, reduce the labor cost, improve the detection accuracy, and provide an effective support basis for further insulator discharge diagnosis. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0041] Figure 1 A flowchart of a model fusion-based insulator discharge defect adaptive recognition method provided by the embodiments of the present application is shown in the figure.

[0042] Figure 2 A working flowchart of obtaining the length of the inscribed polygon of the discharge area provided by the embodiments of the present application is shown in the figure.

[0043] Figure 3 A length data diagram of obtaining the inscribed polygon of the discharge area provided by the embodiments of the present application is shown in the figure.

[0044] Figure 4 Another length data diagram of obtaining the inscribed polygon of the discharge area provided by the embodiments of the present application is shown in the figure.

[0045] Figure 5 A structure block diagram of a model fusion-based insulator discharge defect adaptive recognition system embodiment provided by the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0046] In order to make the objectives, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application in combination with the accompanying drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application, and are not used to limit the present application. In addition, the technical features involved in each embodiment of the present application described below can be combined with each other as long as they do not conflict with each other.

[0047] The terms "first", "second", "third", etc. in the specification and claims of the present application and the above drawings are used to distinguish different objects, and are not used to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or device.

[0048] In one embodiment, as shown in Figure 1 A flowchart of a model fusion-based insulator discharge defect adaptive recognition method provided by an embodiment of the present application is shown in FIG. 1. The method includes the following steps.

[0049] In step 110, picture data set of insulators is obtained, first image processing is performed on the picture data set, first sample set is obtained, and first target detection model is constructed and generated according to the first sample set. The first target detection model is a single-stage target detection model.

[0050] Preferably, the insulator discharge video obtained by the unmanned aerial vehicle cruising is frame-by-frame decomposed to obtain picture data set data1 containing insulators at different angles. First image processing is performed on the picture data set data1. Preferably, the first image processing can include, for example, using rectangular frame labeling software such as Labelme software to label the insulator region in the insulator image in the picture data set data1; based on the position information of the picture data set data1, label text set label1 corresponding to the insulator image is formed; in order to improve the generalization ability of the model, the insulator image is subjected to operations such as filtering, scaling, pixel transformation, etc. to obtain a new data set to join data1 to obtain picture data set data2, and the label text set content is transformed based on the coordinate system before and after the transformation to obtain label text set label2, wherein the label text set label2 does not contain the inclination angle.

[0051] Further, the picture dataset data2 and the label text dataset label2 are input into the single-stage target detection model as a first sample set to train and generate a first target detection model, preferably, the single-stage target detection model selects, for example, a YOLOV5 model. The first sample set is detected to observe the horizontal and vertical insulator detection effect of model inference. According to the actual test result, the model is continuously adjusted. Specifically, the adjustment includes, for example, cleaning of training samples, label correction, model hyperparameter adjustment, iteration of network models and the like; the YOLOV5 model is adjusted to stably detect the insulators in the horizontal and vertical directions, and the first target detection model is obtained.

[0052] It is worth noting that the single-stage target detection model used in the present application can accurately and quickly locate the insulator, has high accuracy and robustness, can realize real-time detection, and is very useful for application scenarios that require fast response. Compared with the horizontal and vertical insulators obtained by the multi-target detection model, the single-target detection model has higher speed and less computing resource demand, and therefore can further improve the detection efficiency of the insulator.

[0053] In step 120, the picture dataset is subjected to second image processing to obtain a second sample set, and a second target detection model is constructed and generated according to the second sample set, wherein the second target detection model is a two-stage target detection model.

[0054] The picture dataset data1 is subjected to second image processing, wherein the second image processing can include, for example, rotating the picture dataset data1 by 90°, 180° and 270° to obtain a picture dataset data3; and using a labeling software such as RoLabelImg software to label the picture dataset data3 to obtain a label text dataset label3, wherein the label text dataset label3 contains an inclination angle.

[0055] Further, the picture dataset data3 and the label text dataset label3 are input into the two-stage target detection model as a second sample set to train and generate a second target detection model, preferably, the two-stage target detection model selects, for example, a S2A-Net model, and the second sample set is detected to observe the horizontal, vertical and inclined insulator detection effect of model inference. According to the actual test result, the model is continuously adjusted. Specifically, the adjustment includes, for example, cleaning of training samples, label correction, model hyperparameter adjustment, iteration of network models and the like; the S2A-Net model is adjusted to stably detect the insulator in the inclined direction, and the second target detection model is obtained.

[0056] Step 130, the picture data of the insulator to be detected under the same acquisition means as the picture data set of the insulator is input to the first target detection model to obtain first coordinate data, and is input to the second target detection model to obtain second coordinate data; and the state information of the insulator to be detected is determined by analyzing the second coordinate data.

[0057] Preferably, the picture data of the insulator containing different angle detection obtained by the unmanned aerial vehicle cruising is obtained, the picture data to be detected is input to the first target detection model to obtain first coordinate data, and is input to the second target detection model to obtain second coordinate data; wherein the output of the second target detection model S2A-Net model also contains angle information; the inclination angle angle of the insulator to be detected is obtained by coordinate calculation, the inclination angle angle of the insulator to be detected is judged, and state classification is performed to obtain state classification results, and the state classification results include the first state and the second state of the insulator to be detected.

[0058] Preferably, the inclination angle angle of the insulator to be detected is judged, for example, if the inclination angle of the insulator to be detected is calculated as belonging to 0°≤|angle|≤2° in the coordinate axis of the second coordinate data, the insulator is judged as horizontal; if 88°≤|angle|≤92°, the insulator is judged as vertical; otherwise, the insulator is judged as inclined; and state classification is performed, for example, the first state includes the insulator judged as horizontal or vertical, and the second state includes the inclination.

[0059] Step 140, the first coordinate data and the second coordinate data are fused and analyzed to obtain the fusion coordinate data of the insulator to be detected, and the length data of the insulator to be detected is calculated according to the fusion coordinate data.

[0060] Preferably, if the state classification result of the insulator to be detected is the first state, the first coordinate data is taken as the fusion coordinate data of the insulator to be detected.

[0061] If the state classification result of the insulator to be detected is the second state, the first coordinate data and the second coordinate data are calculated by intersection and union ratio to obtain the coordinate data with the minimum intersection and union ratio as the fusion coordinate data.

[0062] According to specific needs, different state classification results can be set, in a preferred embodiment of the present application, the first state includes the insulator judged as horizontal or vertical, and the second state includes the inclination.

[0063] Specifically, in a preferred embodiment, a picture data containing three insulators is selected, respectively marked as a, b, c, after the picture is inferred by the first target detection model and the second target detection model, the two models will respectively output 3 rectangular boxes (a1, b1, c1 and a2, b2, c2), the four vertices of each rectangular box in (a1, b1, c1) are the first coordinate data of the insulator, and the four vertices of each rectangular box in (a2, b2, c2) are the second coordinate data of the insulator. According to the second coordinate data (a2, b2, c2), the state information of the insulator to be detected is determined, and it is obtained that a and b belong to the first state of vertical or horizontal, and c belongs to the second state of inclined.

[0064] The first coordinate data (a1, b1) of a and b belonging to the first state is taken as the fusion coordinate data of the insulator to be detected.

[0065] The first coordinate data and the second coordinate data of c belonging to the second state are calculated by intersection and union ratio, wherein the intersection and union ratio of c2 (inclined box) and (a1, b1, c1) is calculated respectively, so that c1 is predicted to be an inclined insulator (the maximum intersection and union ratio is ), and c2 is used to replace c1 to obtain the accurate coordinate data (a1, b1, c2) of the insulator.

[0066] In step 150, the external polygon contour of the discharge area 1 domain in the picture data of the insulator to be detected is determined, and the first length of the discharge area of the insulator to be detected is obtained by calculating the external polygon contour and the fusion coordinate data.

[0067] The picture data of the insulator to be detected is preprocessed by, for example, HSV color model, filtering, and morphological operation to extract the ultraviolet discharge area, and the discharge area of the insulator to be detected is obtained. It should be noted that the acquisition of the ultraviolet discharge area is a technology known to those skilled in the art, and therefore the present application will not be described here.

[0068] Preferably, as shown in the figure, Figure 2 A working flow diagram for obtaining the length of the external polygon of the discharge area is provided for the embodiments of the present application; and the discharge area is surrounded by a polygon approximation method to obtain the external polygon contour of the discharge area; wherein the external polygon contour includes the maximum external rectangle and the maximum external circle.

[0069] The step of calculating the first length of the discharge area of the insulator to be detected according to the external polygon contour and the fusion coordinate data.

[0070] Preferably, the step includes the following cases:

[0071] (1) the state classification result of the insulator to be detected is the first state,

[0072] If the state classification result of the detected insulator is horizontal or vertical, the difference between the maximum and minimum coordinates of the circumscribed polygon contour on the first direction axis or the second direction axis is calculated as the first length.

[0073] Specifically, when the insulator is vertical or horizontal, the maximum value and the minimum value of the polygon contour along the Y axis or the X axis in the same coordinate axis as the second coordinate data are calculated, and the difference is obtained as the first length L of the discharge area contours .

[0074] (2) the state classification result of the insulator to be detected is the second state,

[0075] If the state classification result of the detected insulator is inclined, the first maximum length along the first direction axis and the second maximum length along the second direction axis are determined, the first direction axis and the second direction axis are perpendicular; the first maximum length is the difference between the maximum and minimum coordinates of the circumscribed polygon contour on the first direction axis; the second maximum length is the difference between the maximum and minimum coordinates of the circumscribed polygon contour on the second direction axis;

[0076] When the first maximum length is greater than or equal to the second maximum length, it is determined that the polygon contour is a maximum circumscribed rectangle, and the length of the cutting line of the maximum circumscribed rectangle parallel to the insulator is calculated as the first length;

[0077] When the second maximum length is greater than or equal to the first maximum length, it is determined that the polygon contour is a maximum circumscribed circle, and the diameter of the maximum circumscribed circle is calculated as the first length.

[0078] Specifically, as Figure 3 and 4 the length data of the circumscribed polygon of the discharge area provided by the embodiment of the application is shown in the figure; when the insulator is inclined, the maximum length L x and the maximum length L y on the X axis and the Y axis in the same coordinate axis as the second coordinate data need to be determined; when L x < L y , the discharge length L rec obtained by using the maximum circumscribed rectangle is greater than the discharge length L circle obtained by using the maximum circumscribed circle, which is closer to the real discharge length of the insulator, so the cutting line parallel to the insulator of the maximum circumscribed rectangle is calculated as the first length L contours of the discharge area of the insulator to be detected.

[0079] When L x ≥ L yAt this time, the discharge length L is obtained by using the maximum circumscribed circle circle greater than the discharge length L obtained by using the maximum circumscribed rectangle rec , and is closer to the actual discharge length of the insulator, so that the diameter data of the maximum circumscribed circle is obtained as the first length L of the discharge region of the insulator to be detected contours .

[0080] The length data of the discharge region obtained by selecting different circumscribed rectangular frames according to different inclination angles of different insulators is closer to the actual length of the discharge region, further reduces the error caused by subjective judgment of the test personnel, and the influence of the error of the coordinate data of the insulator obtained by the target detection model on the detection result. The length data of the outer polygon of the discharge region and the length of the ultraviolet discharge region are improved, so that the detection result is more referential, and important support is provided for further insulator discharge defect detection.

[0081] In step 160, a first length in the same direction as the insulator to be detected is selected from the circumscribed polygon contour, and the length ratio of the discharge region on the surface of the insulator to be detected is calculated based on the first length and the length data.

[0082] Preferably, the length ratio of the discharge region on the surface of the insulator to be detected is calculated based on the first length and the length data according to the following formula:

[0083]

[0084] wherein L contours is the first length of the discharge region of the insulator to be detected, L jyz is the length of the target insulator, and is obtained by fusing the coordinate data of the insulator to be detected.

[0085] Further, the discharge defect risk level is judged according to the length ratio of the ultraviolet image discharge region. In one embodiment, when the length ratio Ratio<20, it is judged as low risk; when the length ratio 20≤Ratio<33, it is judged as medium risk; and when the length ratio Ratio≥33, it is judged as high risk. The discharge defect risk level judgment value selected by the present application is only an example, and the discharge defect risk level judgment value can be increased or deleted according to the needs of those skilled in the art.

[0086] Referring to Figure 5As shown in FIG. 1, a structural block diagram of an embodiment of an insulator discharge defect adaptive recognition system based on model fusion according to the present application comprises: a first image processing unit: used for acquiring a picture data set of an insulator, performing first image processing on the picture data set, obtaining a first sample set, and constructing a first target detection model according to the first sample set, wherein the first target detection model is a single-stage target detection model.

[0087] A second image processing unit: used for performing second image processing on the picture data set, obtaining a second sample set, and constructing a second target detection model according to the second sample set, wherein the second target detection model is a double-stage target detection model.

[0088] A picture processing unit: used for inputting picture data of an insulator to be detected under the same acquisition means as the picture data set of the insulator into the first target detection model to obtain first coordinate data, and into the second target detection model to obtain second coordinate data, and analyzing the second coordinate data to determine state information of the insulator to be detected.

[0089] A fusion data analysis unit: used for performing fusion analysis on the first coordinate data and the second coordinate data to obtain fusion coordinate data of the insulator to be detected, and calculating length data of the insulator to be detected according to the fusion coordinate data.

[0090] A discharge region analysis unit: used for determining an external polygon contour of a discharge region in ultraviolet picture data of the picture data of the insulator to be detected, and calculating a first length of the discharge region of the insulator to be detected according to the external polygon contour and the fusion coordinate data.

[0091] An evaluation unit: used for selecting a first length in the external polygon contour which is the same as the direction of the insulator to be detected, and calculating a length proportion of the discharge region on the surface of the insulator to be detected according to the first length and the length data.

[0092] The implementation principle and technical effects of the system are similar to those of the above method, and will not be described here.

[0093] It should be noted that, for the above-mentioned method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited by the action sequence described, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0094] In the above embodiments, the description of each embodiment focuses on different aspects, and the parts not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0095] In several embodiments provided in the present application, it should be understood that the disclosed apparatus can be implemented by other means. For example, the apparatus embodiments described above are only illustrative, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some service interface, apparatus or unit, which can be electrical or other forms.

[0096] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place or distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0097] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0098] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable memory. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product, which is stored in a memory and includes a plurality of instructions for making a computer device (which can be a personal computer, a server or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: a U disk, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0099] Those skilled in the art can understand that all or part of the steps of various methods in the above embodiments can be instructed by a program to relevant hardware, and the program can be stored in a computer readable memory, which can include a flash disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0100] The above merely provides illustrative examples of the present disclosure, but cannot limit the scope of the present disclosure. Any equivalent changes and modifications made according to the teachings of the present disclosure shall fall within the scope of the present disclosure. Those skilled in the art will readily understand the embodiments of the present disclosure upon considering the specification and practicing the disclosure herein. The present application is intended to cover any variations, uses, or adaptive changes of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and examples are merely regarded as illustrative, and the scope and spirit of the present disclosure are defined by the claims.

[0101] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present disclosure.

[0102] Those skilled in the art will readily understand that the above merely provides preferred embodiments of the present application, but is not used to limit the present application, and any modifications, equivalent replacements and improvements made within the spirit and principle of the present application shall fall within the scope of the present application.

Claims

1. A model fusion-based insulator discharge defect adaptive recognition method, characterized in that, The method comprises the following steps: Obtaining a picture data set of the insulator, performing first image processing on the picture data set to obtain a first sample set, and constructing a first target detection model according to the first sample set, wherein the first target detection model is a single-stage target detection model; Performing second image processing on the picture data set to obtain a second sample set, and constructing a second target detection model according to the second sample set, wherein the second target detection model is a two-stage target detection model; Inputting picture data of the insulator to be detected under the same collection means as the picture data set of the insulator into the first target detection model to obtain first coordinate data; and inputting the picture data into the second target detection model to obtain second coordinate data; analyzing the second coordinate data, calculating the second coordinate data to obtain the inclination angle of the insulator to be detected, and classifying the state to obtain a state classification result, wherein the state classification result comprises a first state and a second state of the insulator to be detected; the first state is a vertical or horizontal state under the same marking means as the second coordinate data; and the second state is an inclined state under the same marking means as the second coordinate data; Performing fusion analysis on the first coordinate data and the second coordinate data to obtain fusion coordinate data of the insulator to be detected; if the state classification result of the insulator to be detected is the first state, the first coordinate data is taken as the fusion coordinate data of the insulator to be detected; If the state classification result of the insulator to be detected is the second state, the first coordinate data and the second coordinate data are calculated by intersection over union to obtain the second coordinate data with the maximum intersection over union as the fusion coordinate data; and calculating the length data of the insulator to be detected according to the fusion coordinate data; Determining the external polygon contour of the discharge region in the ultraviolet picture data of the picture data of the insulator to be detected; and calculating the first length of the discharge region of the insulator to be detected according to the external polygon contour and the fusion coordinate data; Selecting the first length in the external polygon contour which is the same as the direction of the insulator to be detected, and calculating the length ratio of the discharge region on the surface of the insulator to be detected according to the first length and the length data.

2. The model fusion-based insulator discharge defect adaptive recognition method according to claim 1, characterized in that, The step of determining the external polygon contour of the discharge region in the ultraviolet picture data of the picture data of the insulator to be detected comprises: Determining the discharge region in the ultraviolet picture data of the picture data of the insulator to be detected, and obtaining the external polygon contour of the discharge region by using the polygon approximation method; wherein the external polygon contour comprises a maximum external rectangle and a maximum external circle.

3. The model fusion-based insulator discharge defect adaptive recognition method according to claim 2, characterized in that, The step of calculating the first length of the discharge region of the insulator to be detected according to the external polygon contour and the fusion coordinate data comprises: If the state classification result of the insulator to be detected is the first state, the difference between the maximum and minimum coordinates of the external polygon contour in the first direction axis or the second direction axis is calculated as the first length.

4. The model fusion-based insulator discharge defect adaptive recognition method according to claim 2, characterized in that, The step of calculating the first length of the discharge region of the insulator to be detected according to the circumscribed polygon contour and the fusion coordinate data further comprises: If the state classification result of the insulator to be detected is the second state, the first maximum length along a first direction axis and the second maximum length along a second direction axis are determined, the first direction axis and the second direction axis are perpendicular; the first maximum length is the difference between the maximum and minimum coordinates of the circumscribed polygon contour on the first direction axis; the second maximum length is the difference between the maximum and minimum coordinates of the circumscribed polygon contour on the second direction axis; When the first maximum length is greater than or equal to the second maximum length, it is determined that the polygon contour is a maximum circumscribed rectangle, and the length of the cutting line of the maximum circumscribed rectangle parallel to the insulator is calculated as the first length; When the second maximum length is greater than or equal to the first maximum length, it is determined that the polygon contour is a maximum circumscribed circle, and the diameter of the maximum circumscribed circle is calculated as the first length.

5. The model fusion-based insulator discharge defect adaptive recognition method according to claim 1, characterized in that, The calculation formula of the length proportion is as follows: wherein, is a length of the target insulator, is a first length of a discharge region of the insulator to be detected.

6. A model fusion-based insulator discharge defect adaptive recognition system, characterized in that, It comprises: A first image processing unit is configured to obtain a picture data set of an insulator, perform first image processing on the picture data set, obtain a first sample set, and construct a first target detection model according to the first sample set, wherein the first target detection model is a single-stage target detection model; A second image processing unit is configured to perform second image processing on the picture data set, obtain a second sample set, and construct a second target detection model according to the second sample set, wherein the second target detection model is a two-stage target detection model; A picture processing unit is configured to input picture data of an insulator to be detected, which is collected by the same means as the picture data set of the insulator, into the first target detection model to obtain first coordinate data, and into the second target detection model to obtain second coordinate data; analyze the second coordinate data, calculate the inclination angle of the insulator to be detected according to the second coordinate data, and perform state classification to obtain a state classification result, wherein the state classification result comprises a first state and a second state of the insulator to be detected; the first state is a vertical or horizontal state under the same marking means as the second coordinate data; and the second state is an inclined state under the same marking means as the second coordinate data; A fusion data analysis unit is configured to perform fusion analysis on the first coordinate data and the second coordinate data to obtain fusion coordinate data of the insulator to be detected; if the state classification result of the insulator to be detected is the first state, the first coordinate data is taken as the fusion coordinate data of the insulator to be detected; If the state classification result of the insulator to be detected is the second state, the first coordinate data and the second coordinate data are subjected to intersection and union ratio calculation, the second coordinate data with the maximum intersection and union ratio is taken as the fusion coordinate data, and the length data of the insulator to be detected is calculated according to the fusion coordinate data. ​ The discharge region analysis unit is configured to determine an outer polygon contour of a discharge region extracted from ultraviolet image data in the image data of the insulator to be detected, and calculate the first length of the discharge region of the insulator to be detected according to the outer polygon contour and the fusion coordinate data. The evaluation unit is configured to select a first length in the outer polygon contour which is the same as the direction of the insulator to be detected, and calculate the length ratio of the discharge region on the surface of the insulator to be detected according to the first length and the length data.

7. The model fusion based insulator discharge defect adaptive recognition system of claim 6, wherein, The calculation formula of the length ratio calculated by the evaluation unit is as follows: wherein, is a first length of a discharge region of the insulator to be detected, is a length of the target insulator.

Citation Information

Patent Citations

  • GIS equipment fault judgment method based on machine learning

    CN110888025A

  • Method for detecting discharge distribution of zero-value insulator by adopting GaN red and ultraviolet sensor

    CN113808109A