Hydrophobicity evaluation method and system based on composite insulator umbrella skirt surface water drop group

Abstract

The invention provides a hydrophobicity evaluation method and system based on a composite insulator umbrella skirt surface water drop group, and relates to the field of insulation material hydrophobicity analysis, and the method comprises the steps: constructing a composite insulator hydrophobicity evaluation model based on a Faster R-CNN network and a Fourier model; collecting a water drop group image; adopting a structured light projection method to pre-process the water drop group image, and constructing a data set; training a composite insulator hydrophobicity evaluation model through the data set; obtaining a three-dimensional morphology graph of an umbrella skirt water drop group of the to-be-detected insulator; and inputting the three-dimensional morphology graph into the trained composite insulator hydrophobicity evaluation model for classification to obtain the hydrophobicity degree grade of the insulator to be detected. The composite insulator hydrophobicity evaluation model can more effectively extract the image features in the external environment, thereby improving the accuracy of hydrophobicity evaluation.

Description

Technical Field

[0001] The present application relates to the field of hydrophobicity analysis of insulating materials, and in particular to a hydrophobicity evaluation method and system based on water droplets on the surface of composite insulator sheds. Background Art

[0002] Composite insulators have the characteristics of light weight, strong anti-pollution, good hydrophobicity and long service life. my country began to gradually promote the use of composite insulators in the 1980s. Today, composite insulators have been widely used in high-voltage transmission power grids. According to relevant data, the number of composite insulators in use in my country has exceeded five million. The practice of power systems shows that with the increase of operating time, the hydrophobicity of composite insulator materials is more serious. The evaluation of the hydrophobicity of composite insulators is an important topic in power system maintenance. This method uses the distribution characteristics of water droplets on the surface of composite insulator sheds to evaluate their hydrophobicity. Since composite insulators are exposed to various environmental factors during operation, the hydrophobicity of the surface of the sheds will gradually decrease over time, which directly affects the insulation performance and hydrophobicity of the insulators.

[0003] By analyzing the shape, size and distribution of water droplets on the surface of the insulator shed, and then judging the level of hydrophobicity of the insulator shed surface, its hydrophobicity state can be effectively evaluated. This technology not only improves the accuracy of hydrophobicity evaluation, but also provides a new idea for preventing power system failures. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for evaluating the hydrophobicity of water droplets on the surface of a composite insulator shed in order to solve the problem that the existing method for evaluating the hydrophobicity of water droplets on the surface of a composite insulator shed cannot take into account both detection efficiency and detection accuracy.

[0005] The above-mentioned purpose of the present application is achieved through the following technical solutions: S1: A composite insulator hydrophobicity evaluation model constructed based on the Faster R-CNN network and Fourier model; S2: Collect water drop group images; use structured light projection method to preprocess the water drop group images and construct a data set; S3: The composite insulator hydrophobicity evaluation model is trained using the dataset; S4: obtaining a three-dimensional topographic image of a water droplet group on the shed of the insulator to be tested; S5: Input the three-dimensional morphology image into the trained composite insulator hydrophobicity evaluation model for classification, and obtain the hydrophobicity level of the insulator to be tested.

[0006] Optionally, step S1 includes: replacing the VGG16 skeleton network in the Faster R-CNN network with a ResNet50 network.

[0007] Optionally, step S2 includes: S21: performing water spraying treatment on the shed surfaces of composite insulators with different hydrophobicity levels to form different water drop groups on the shed surfaces, and obtaining water drop group images of the insulators; S22: Classify the water drop group image through the Faster R-CNN network to determine the hydrophobicity level label of the water drop group image; S22: Reconstruct the water drop group image using Fourier transform profilometry in the structured light projection method; S23: Through image processing methods such as horizontal flipping, vertical flipping and rotation, data enhancement is performed on the reconstructed water drop group image to construct a data set.

[0008] Optionally, step S22 includes: S22a: graying the water drop group image; S22b: using bilinear interpolation, adjusting the size of the grayscale image; S22c: Classify the water droplets in the grayscale image through the Faster R-CNN network to obtain the local feature parameters of the water droplet distribution; Based on digital image processing, the global parameters about the bright spot area and morphology of water droplets are extracted from the grayscale image; S22d: Determine the water drop distribution characteristic parameters through global parameters and local characteristic parameters; S22e: Classify the water droplets according to the characteristic parameters of the water droplet distribution, and determine the hydrophobicity level label corresponding to the water droplet group image.

[0009] Optionally, step S4 includes: S41: obtaining a deformed grating image of a water droplet group on the shed of the insulator to be inspected; S42: determining a two-dimensional conventional image of the insulator to be inspected according to the deformed grating image; S43: Based on histogram equalization and the morphological characteristics of the insulator to be inspected, a preliminary judgment is made on the aging inspection area of ​​the two-dimensional traditional image: If it is determined that there is no problem with the aging area to be inspected, a preliminary inspection will be carried out on the next area; If it is determined that there is a problem with the aging area to be inspected, the three-dimensional morphology of the composite insulator is reconstructed based on Fourier transform profilometry, and combined with the deformed grating image, a three-dimensional morphology image of the shed water droplet group to be inspected is obtained.

[0010] Optionally, step S5 includes: S51: Based on a local weighted regression algorithm, a surface fitting is performed on the area to be inspected of the three-dimensional topography image to obtain a surface fitting result; S52: subtracting the three-dimensional result of the three-dimensional topography image from the surface fitting result to calculate a root mean square error; judging the degree of inspection of the area to be inspected according to the root mean square error; S53: If the degree to be inspected is greater than a preset threshold, the area to be inspected is classified by the composite insulator hydrophobicity evaluation model, as follows: Based on the Fourier model, after preprocessing the areas to be inspected whose inspection degree is greater than the preset threshold; The pre-processed three-dimensional topography image of the area to be inspected is uniformly cut into M segments to obtain M cross-sectional images and draw M three-dimensional images of water drops; The number of water droplets N in each three-dimensional water droplet image is counted to obtain the hydrophobicity grade; The hydrophobicity levels of the M cross-sectional images are averaged to obtain the hydrophobicity level of the insulator to be tested.

[0011] A hydrophobicity evaluation system based on water droplets on the surface of composite insulator sheds, the system comprising: an image acquisition optical axis, a charge coupled device, a computer, and a display; The image acquisition optical axis is connected to a charge coupled device, the charge coupled device is connected to a computer; the computer is connected to a display; The image acquisition optical axis is used to acquire the deformed grating image of the insulator; The charge coupled device is used to convert the deformed grating image into a three-dimensional topography image; The computer is used to construct a composite insulator hydrophobicity evaluation model based on the Faster R-CNN network and the Fourier model; The computer is also used to collect water drop group images; the water drop group images are preprocessed using a structured light projection method to construct a data set; The computer is also used to train a composite insulator hydrophobicity evaluation model through a data set; The computer is also used to input the three-dimensional topography image into the trained composite insulator hydrophobicity evaluation model for classification, so as to obtain the hydrophobicity level of the insulator to be tested; The display is used to visualize the three-dimensional topography and the hydrophobicity level.

[0012] A computer-readable storage medium stores instructions. When the instructions are executed, a method for evaluating the hydrophobicity of water droplets on the surface of a composite insulator shed is executed.

[0013] The beneficial effects of the technical solution provided by this application are: 1. ResNet50 is used as a feature extraction network to replace the skeleton network VGG16 in the traditional Faster R-CNN network to solve the problem of gradient disappearance that may be caused by multiple inter-layer propagation and improve the detection accuracy of the model.

[0014] 2. By spraying water and taking pictures on the insulator, it is more convenient to extract sample data. And because a large amount of insulator data can be obtained, the representativeness of the sample is relatively improved, and the accumulation of hydrophobicity data increases every year, which also facilitates the optimization of subsequent hydrophobicity evaluation indicators. The data samples used in the training model of this application are highly representative, the sample data are easy to extract, and the data samples are rich. Compared with the existing insulator hydrophobicity evaluation method, the method of sampling and offline sample detection has been greatly improved, which improves the prediction accuracy of the model.

[0015] 3. When detecting water droplets on the surface of the composite insulator shed, the detection efficiency is improved by making a preliminary judgment on the aging inspection area of ​​the two-dimensional traditional image based on histogram equalization and the morphological characteristics of the insulator to be detected, as well as judging the degree of inspection of the inspection area to be detected. Compared with the existing insulator hydrophobicity evaluation method of sampling and offline sample detection, it has stronger real-time performance and higher detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present application will be further described below with reference to the accompanying drawings and embodiments, in which: Figure 1 It is a step diagram of a method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds in an embodiment of the present application; Figure 2 is an oblique topographic image of a water droplet group on an insulator based on a method for evaluating the hydrophobicity of a water droplet group on the surface of a composite insulator shed in an embodiment of the present application; Figure 3 is a three-dimensional morphology diagram of a water droplet group on an insulator based on a method for evaluating the hydrophobicity of a water droplet group on the surface of a composite insulator shed in an embodiment of the present application; Figure 4 It is a plan view of a water drop group on an insulator shed based on a method for evaluating the hydrophobicity of water drop groups on the surface of a composite insulator shed in an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to have a clearer understanding of the technical features, purposes and effects of the present application, the specific implementation methods of the present application are now described in detail with reference to the accompanying drawings.

[0018] The embodiments of the present application provide a method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds.

[0019] Please refer to Figure 1 , Figure 1This is a step diagram of a method for evaluating the hydrophobicity of water droplets on the surface of a composite insulator shed in an embodiment of the present application, comprising: S1: A composite insulator hydrophobicity evaluation model constructed based on the Faster R-CNN network and Fourier model; Step S1 includes: replacing the VGG16 skeleton network in the Faster R-CNN network with the ResNet50 network.

[0020] Specifically, considering the high difficulty of classifying water drops / water marks on insulators, this paper uses ResNet50 as the feature extraction network to replace the traditional skeleton network VGG16. The ResNet network solves the problem that in deep networks, the gradient may disappear due to multiple inter-layer propagation.

[0021] S2: Collect water drop group images; use structured light projection method to preprocess the water drop group images and construct a data set; Step S2 includes: S21: performing water spraying treatment on the shed surfaces of composite insulators with different hydrophobicity levels to form different water drop groups on the shed surfaces, and obtaining water drop group images of the insulators; S22: Classify the water drop group image through the Faster R-CNN network to determine the hydrophobicity level label of the water drop group image; S22: Reconstruct the water drop group image using Fourier transform profilometry in the structured light projection method; Step S22 includes: S22a: graying the water drop group image; Specifically, image graying eliminates redundant information and converts color images into grayscale images.

[0022] S22b: using bilinear interpolation, adjusting the size of the grayscale image; S22c: Classify the water droplets in the grayscale image through the Faster R-CNN network to obtain the local feature parameters of the water droplet distribution; Based on digital image processing, the global parameters about the bright spot area and morphology of water droplets are extracted from the grayscale image; Specifically, the local characteristic parameters of water drop distribution include: the number of detected water drops / water marks of each level, the maximum detected area, the total detected area, and the average detected area. The area is replaced by the detection frame area (pixel).

[0023] Specifically, based on the global parameters of water droplet distribution characteristics obtained through digital image processing, the preprocessing combines the local parameters of water droplet distribution characteristics formed by the target detection algorithm with the global feature parameters currently extracted based on digital image processing to enhance the parameter set's ability to characterize the hydrophobicity of the image.

[0024] S22d: Determine the water drop distribution characteristic parameters through global parameters and local characteristic parameters; S22e: Classify the water droplets according to the characteristic parameters of the water droplet distribution, and determine the hydrophobicity level label corresponding to the water droplet group image.

[0025] S23: Through image processing methods such as horizontal flipping, vertical flipping and rotation, data enhancement is performed on the reconstructed water drop group image to construct a data set.

[0026] Specifically, the number of water drop samples of five levels can meet the requirements of neural network training, but the number of L5 water trace samples is relatively small. During the training, three data enhancement methods, namely horizontal flipping, vertical flipping and 90° rotation, were used to expand the data capacity to three times the original.

[0027] S3: The composite insulator hydrophobicity evaluation model is trained using the dataset; S4: obtaining a three-dimensional topographic image of a water droplet group on the shed of the insulator to be tested; Step S4 includes: S41: obtaining a deformed grating image of a water droplet group on the shed of the insulator to be inspected; Specifically, the deformed grating image is as follows:

[0028] in, Represents the three-dimensional coordinate function of the grating image; is the non-uniform reflectivity of the shed surface; Indicates the horizontal and vertical coordinates; Indicates amplitude adjustment; represents the number of Fourier transforms; represents the natural exponential function; is the fundamental frequency of the deformed grating image, It indicates the phase modulation caused by the ups and downs of the surface of the object's shed skirt; It represents the phase modulation caused by the ups and downs of the object's shed surface.

[0029] S42: determining a two-dimensional conventional image of the insulator to be inspected according to the deformed grating image; S43: Based on histogram equalization and the morphological characteristics of the insulator to be inspected, a preliminary judgment is made on the aging inspection area of ​​the two-dimensional traditional image: If it is determined that there is no problem with the aging area to be inspected, a preliminary inspection will be carried out on the next area; If it is determined that there is a problem with the aging area to be inspected, the three-dimensional morphology of the composite insulator is reconstructed based on Fourier transform profilometry, and combined with the deformed grating image, a three-dimensional morphology image of the shed water droplet group to be inspected is obtained.

[0030] Specifically, the steps of reconstructing the three-dimensional morphology of the composite insulator based on Fourier transform profilometry and combining the deformed grating image to obtain the three-dimensional morphology image of the shed water drop group to be detected specifically include: The deformed grating image is subjected to Fourier transform, spectrum filtering and inverse Fourier transform, and the light field distribution of the deformed grating image in the spatial domain is expressed as:

[0031] After the image acquisition optical axis collects the deformed grating image, it transmits the image to the ground processing equipment via wireless transmission. The equipment performs Fourier transform on the deformed grating image, performs spectrum filtering, extracts the base frequency component shown by the shaded image, and then performs inverse Fourier transform on it.

[0032] When the height of each point on the shed surface When it is 0, that is, when measuring the reference plane, the pre-stored grating image is expressed as:

[0033] Perform Fourier transform, spectrum filtering, and inverse Fourier transform on the pre-stored grating image to obtain:

[0034] in ; It is the additional phase modulation introduced by the divergent illumination due to the exit pupil of the projection system being at a finite distance. If the exit pupil of the projection system is at infinity, the additional phase is zero.

[0035] For the case of diffuse illumination, the phase modulation caused by height alone for:

[0036] in Represents the vector length between two adjacent points; Get phase modulation:

[0037] In the formula Indicates taking the imaginary part of a complex number, and * indicates conjugation; After phase unwrapping, the height values ​​of each point on the shed surface are obtained according to the similarity relationship between triangle HCD and HEpEc. for:

[0038] According to the height values ​​of each point on the shed surface, a three-dimensional morphology map of the shed water droplet group of the insulator to be tested is constructed.

[0039] S5: Input the three-dimensional morphology image into the trained composite insulator hydrophobicity evaluation model for classification, and obtain the hydrophobicity level of the insulator to be tested.

[0040] Step S5 includes: S51: Based on a local weighted regression algorithm, a surface fitting is performed on the area to be inspected of the three-dimensional topography image to obtain a surface fitting result; S52: subtracting the three-dimensional result of the three-dimensional topography image from the surface fitting result to calculate a root mean square error; judging the degree of inspection of the area to be inspected according to the root mean square error; S53: If the degree to be inspected is greater than a preset threshold, the area to be inspected is classified by the composite insulator hydrophobicity evaluation model, as follows: Based on the Fourier model, after preprocessing the areas to be inspected whose inspection degree is greater than the preset threshold; The pre-processed three-dimensional topography image of the area to be inspected is uniformly cut into M segments to obtain M cross-sectional images and draw M three-dimensional images of water drops; The number of water droplets N in each three-dimensional water droplet image is counted to obtain the hydrophobicity grade; The hydrophobicity levels of the M cross-sectional images are averaged to obtain the hydrophobicity level of the insulator to be tested.

[0041] Specifically, the three-dimensional image of water droplets is as follows Figure 2-4 As shown, Figure 2 This is the oblique morphology of the water drop group on the insulator; Figure 3 This is a three-dimensional morphology diagram of the water drop group on the insulator; Figure 4 It is a plan view of the water drop group on the insulator shed.

[0042] A hydrophobicity evaluation system based on water droplets on the surface of composite insulator sheds, the system comprising: an image acquisition optical axis, a charge coupled device, a computer, and a display; The image acquisition optical axis is connected to a charge coupled device, the charge coupled device is connected to a computer; the computer is connected to a display; The image acquisition optical axis is used to acquire the deformed grating image of the insulator; The charge coupled device is used to convert the deformed grating image into a three-dimensional topography image; The computer is used to construct a composite insulator hydrophobicity evaluation model based on the Faster R-CNN network and the Fourier model; The computer is also used to collect water drop group images; the water drop group images are preprocessed using a structured light projection method to construct a data set; The computer is also used to train a composite insulator hydrophobicity evaluation model using a data set; The computer is also used to input the three-dimensional topography image into the trained composite insulator hydrophobicity evaluation model for classification, so as to obtain the hydrophobicity level of the insulator to be tested; The display is used to visualize the three-dimensional topography and the hydrophobicity level.

[0043] The advantage of the present invention is that by spraying water on insulators through drones and combining with the constructed composite insulator hydrophobicity evaluation model, the hydrophobicity level of composite insulators in outdoor environments can be effectively evaluated, thereby improving the ability to solve the problem that the existing comprehensive hydrophobicity status evaluation system does not have the function of analyzing the hydrophobicity characteristics of insulating materials under different conditions.

[0044] The present application also discloses a computer-readable storage medium storing a plurality of instructions suitable for loading by a processor to execute the above-mentioned method for evaluating the hydrophobicity based on water droplets on the surface of composite insulator sheds.

[0045] The above are only exemplary embodiments of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure.

[0046] This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art not described in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds, characterized in that: The method comprises the following steps: S1: A composite insulator hydrophobicity evaluation model constructed based on the Faster R-CNN network and Fourier model; S2: Collect water drop group images; use structured light projection method to preprocess the water drop group images and construct a data set; S3: The composite insulator hydrophobicity evaluation model is trained using the dataset; S4: obtaining a three-dimensional topographic image of a water droplet group on the shed of the insulator to be tested; S5: Input the three-dimensional morphology image into the trained composite insulator hydrophobicity evaluation model for classification, and obtain the hydrophobicity level of the insulator to be tested.

2. A method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds according to claim 1, characterized in that: Step S1 includes: replacing the VGG16 skeleton network in the Faster R-CNN network with the ResNet50 network.

3. The method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds according to claim 1, characterized in that: Step S2 includes: S21: performing water spraying treatment on the shed surfaces of composite insulators with different hydrophobicity levels to form different water drop groups on the shed surfaces, and obtaining water drop group images of the insulators; S22: Classify the water drop group image through the Faster R-CNN network to determine the hydrophobicity level label of the water drop group image; S22: Reconstruct the water drop group image using Fourier transform profilometry in the structured light projection method; S23: Through image processing methods such as horizontal flipping, vertical flipping and rotation, data enhancement is performed on the reconstructed water drop group image to construct a data set.

4. The method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds according to claim 3, characterized in that: Step S22 includes: S22a: graying the water drop group image; S22b: using bilinear interpolation, adjusting the size of the grayscale image; S22c: Classify the water droplets in the grayscale image through the Faster R-CNN network to obtain the local feature parameters of the water droplet distribution; Based on digital image processing, the global parameters about the bright spot area and morphology of water droplets are extracted from the grayscale image; S22d: Determine the water drop distribution characteristic parameters through global parameters and local characteristic parameters; S22e: Classify the water droplets according to the characteristic parameters of the water droplet distribution, and determine the hydrophobicity level label corresponding to the water droplet group image.

5. The method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds according to claim 1, characterized in that: Step S4 includes: S41: Acquire a deformed grating image of a water droplet group on the shed of the insulator to be inspected; S42: determining a two-dimensional conventional image of the insulator to be inspected according to the deformed grating image; S43: Based on histogram equalization and the morphological characteristics of the insulator to be inspected, a preliminary judgment is made on the aging inspection area of ​​the two-dimensional traditional image: If it is determined that there is no problem with the aging area to be inspected, a preliminary inspection will be carried out on the next area; If it is determined that there is a problem with the aging area to be inspected, the three-dimensional morphology of the composite insulator is reconstructed based on Fourier transform profilometry, and combined with the deformed grating image, a three-dimensional morphology image of the shed water droplet group to be inspected is obtained.

6. The method for evaluating the hydrophobicity of water droplets on the surface of composite insulator sheds according to claim 1, characterized in that: Step S5 includes: S51: Based on a local weighted regression algorithm, a surface fitting is performed on the area to be inspected of the three-dimensional topography image to obtain a surface fitting result; S52: subtracting the three-dimensional result of the three-dimensional topography image from the surface fitting result to calculate a root mean square error; judging the degree of inspection of the area to be inspected according to the root mean square error; S53: If the degree to be inspected is greater than a preset threshold, the area to be inspected is classified by the composite insulator hydrophobicity evaluation model, as follows: Based on the Fourier model, after preprocessing the area to be inspected whose detection degree is greater than the preset threshold; The pre-processed three-dimensional topography image of the area to be inspected is uniformly cut into M segments to obtain M cross-sectional images and draw M three-dimensional images of water drops; The number of water droplets N in each three-dimensional water droplet image is counted to obtain the hydrophobicity grade; The hydrophobicity levels of the M cross-sectional images are averaged to obtain the hydrophobicity level of the insulator to be tested.

7. A hydrophobicity evaluation system based on water droplets on the surface of composite insulator sheds, used to implement a hydrophobicity evaluation method based on water droplets on the surface of composite insulator sheds as claimed in claims 1 to 6, characterized in that: The system comprises: an image acquisition optical axis, a charge coupled device, a computer, and a display; The image acquisition optical axis is connected to a charge coupled device, the charge coupled device is connected to a computer; the computer is connected to a display; The image acquisition optical axis is used to acquire the deformed grating image of the insulator; The charge coupled device is used to convert the deformed grating image into a three-dimensional topography image; The computer is used to construct a composite insulator hydrophobicity evaluation model based on the Faster R-CNN network and the Fourier model; The computer is also used to collect water drop group images; the water drop group images are preprocessed using a structured light projection method to construct a data set; The computer is also used to train a composite insulator hydrophobicity evaluation model using a data set; The computer is also used to input the three-dimensional topography image into the trained composite insulator hydrophobicity evaluation model for classification, so as to obtain the hydrophobicity level of the insulator to be tested; The display is used to visualize the three-dimensional topography and the hydrophobicity level.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed by a computer, the method according to any one of claims 1 to 6 is executed.

Images