Insulation equipment detection method, device, electronic equipment and storage medium

Abstract

The present application discloses a method, device, electronic device and storage medium for detecting insulating equipment, and belongs to the field of intelligent monitoring technology. It includes: acquiring multiple frames of infrared images, and the multiple frames of infrared images are obtained by collecting images of the insulating equipment to be detected and heated. Based on the multiple frames of infrared images, the automatic encoding model is trained to obtain a target automatic encoding model, and the target automatic encoding model can determine the degree of damage to the pixel points in any infrared image. According to the multiple frames of infrared images and the target automatic encoding model, a characteristic image of the insulating equipment is constructed. According to the characteristic image, the damage of the insulating equipment is detected. The detection method based on infrared images of the present application has the characteristic of being able to detect online at a long distance, so it avoids the need for power outage detection of power equipment, thereby avoiding affecting the normal use of the power equipment.

Description

Technical Field

[0001] The present application relates to the field of intelligent monitoring technology, and in particular to a detection method, device, electronic device and storage medium for insulating equipment. Background Art

[0002] Insulation equipment is widely used in transmission lines, substations, converter stations and other scenarios. During the life cycle of insulation equipment, due to inherent defects in the manufacturing process and the influence of environmental factors, internal damage such as cracks, air gaps, and delamination may occur. This may cause insulation failure and even lead to safety accidents.

[0003] Therefore, when evaluating the insulation status of power equipment, it is necessary to detect the damage inside the insulation equipment without damaging the insulation equipment. At present, the detection can generally be carried out through ultrasonic transducers and ultrasonic receivers. For example, ultrasonic transducers can emit ultrasonic waves, and ultrasonic receivers can receive and display the reflected echoes. In this way, the detection result can be determined based on the distortion of the displayed waveform.

[0004] However, when using ultrasonic methods for detection, since high voltage is dangerous, detection can only be performed when the power equipment is powered off. Summary of the invention

[0005] The embodiments of the present application provide a method, device, electronic device and storage medium for detecting insulation equipment, which can solve the problem that the normal use of the power equipment is affected due to the need for power outage detection when detecting insulation equipment. The technical solution is as follows:

[0006] In a first aspect, a method for detecting an insulation device is provided, the method comprising:

[0007] Acquire multiple frames of infrared images, wherein the multiple frames of infrared images are acquired by collecting images of the insulation device to be inspected and after being heated;

[0008] Based on the multiple frames of infrared images, an automatic encoding model is trained to obtain a target automatic encoding model, wherein the target automatic encoding model can determine the degree of damage of a pixel point in any infrared image;

[0009] Constructing a characteristic image of the insulation equipment according to the multiple frames of infrared images and the target automatic encoding model;

[0010] The damage condition of the insulating device is detected according to the characteristic image.

[0011] As an example of the present application, the step of training the automatic encoding model based on the multiple frames of infrared images includes:

[0012] Performing median filtering on each infrared image frame in the multiple infrared image frames;

[0013] Selecting a detection area in each frame of the infrared image after median filtering to obtain a plurality of local area images, wherein the detection area is the area where the insulation equipment is located;

[0014] Based on the multiple local area images, training the auto-encoding model;

[0015] The step of constructing the characteristic image of the insulation equipment according to the multiple frames of infrared images and the target automatic encoding model includes:

[0016] A feature image of the insulation device is constructed according to the multiple local area images and the target automatic encoding model.

[0017] As an example of the present application, the step of training the auto-encoding model based on multiple local area images includes:

[0018] Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area;

[0019] Select k vectors from the m vectors, where k is an integer greater than 1 and less than m;

[0020] Combining the k vectors into a matrix;

[0021] Normalizing the matrix;

[0022] The automatic encoding model is trained based on the k normalized vectors.

[0023] As an example of the present application, constructing the feature image of the insulation device according to the multiple local area images and the target automatic encoding model includes:

[0024] Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area;

[0025] Inputting the m vectors into the target auto-encoding model, and outputting m eigenvalues, each of the m eigenvalues ​​being used to indicate the degree of damage of a corresponding pixel point;

[0026] A characteristic image of the insulation device is constructed based on the m characteristic values.

[0027] As an example of the present application, detecting the damage condition of the insulation device according to the characteristic image includes:

[0028] Performing a binary segmentation process on the feature image to obtain a binary image;

[0029] Determine all connected regions in the binary image;

[0030] The damage condition of the insulation device is detected according to the area of ​​each connected region among all the connected regions.

[0031] As an example of the present application, detecting the damage of the insulating device according to the area of ​​each connected area among all the connected areas includes:

[0032] If there is a connected area whose area is greater than or equal to the area threshold among all the connected areas, it is determined that the insulation device is seriously damaged.

[0033] As an example of the present application, the multiple frames of infrared images include infrared images after the kth frame, where k is an integer greater than 1.

[0034] In a second aspect, a detection device for an insulation device is provided, the device comprising:

[0035] An acquisition module, used for acquiring multiple frames of infrared images, wherein the multiple frames of infrared images are acquired by acquiring images of the insulation device to be inspected and after being heated;

[0036] A training module, used for training the automatic encoding model based on the multiple frames of infrared images to obtain a target automatic encoding model, wherein the target automatic encoding model can determine the degree of damage of a pixel point in any infrared image;

[0037] A construction module, used for constructing a characteristic image of the insulation device according to the multiple frames of infrared images and the target automatic encoding model;

[0038] A detection module is used to detect damage to the insulation device according to the characteristic image.

[0039] As an example of the present application, the training module is used to:

[0040] Performing median filtering on each infrared image frame in the multiple infrared image frames;

[0041] Selecting a detection area in each frame of the infrared image after median filtering to obtain a plurality of local area images, wherein the detection area is the area where the insulation equipment is located;

[0042] Based on the multiple local area images, training the auto-encoding model;

[0043] The construction module constructs a feature image of the insulation device according to the multiple frames of infrared images and the target automatic encoding model, including:

[0044] The construction module constructs a feature image of the insulation device according to the multiple local area images and the target automatic encoding model.

[0045] As an example of the present application, the training module is used to:

[0046] Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area;

[0047] Select k vectors from the m vectors, where k is an integer greater than 1 and less than m;

[0048] Combining the k vectors into a matrix;

[0049] Normalizing the matrix;

[0050] The automatic encoding model is trained based on the k normalized vectors.

[0051] As an example of this application, the building blocks are used to:

[0052] Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area;

[0053] Inputting the m vectors into the target auto-encoding model, and outputting m eigenvalues, each of the m eigenvalues ​​being used to indicate the degree of damage of a corresponding pixel point;

[0054] A characteristic image of the insulation device is constructed based on the m characteristic values.

[0055] As an example of the present application, the detection module is used to:

[0056] Performing a binary segmentation process on the feature image to obtain a binary image;

[0057] Determine all connected regions in the binary image;

[0058] The damage condition of the insulation device is detected according to the area of ​​each connected region among all the connected regions.

[0059] As an example of the present application, the detection module is used to:

[0060] If there is a connected area whose area is greater than or equal to the area threshold among all the connected areas, it is determined that the insulation device is seriously damaged.

[0061] As an example of the present application, the multiple frames of infrared images include infrared images after the kth frame, where k is an integer greater than 1.

[0062] According to a third aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method described in any one of the first aspects is implemented.

[0063] According to a fourth aspect, a computer-readable storage medium is provided, on which instructions are stored. When the instructions are executed by a processor, the method described in any one of the first aspects is implemented.

[0064] In a fifth aspect, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute the method as described in any one of the first aspects.

[0065] The beneficial effects of the technical solution provided by the embodiment of the present application are:

[0066] Acquire multiple frames of infrared images, which are obtained by collecting images of the insulating equipment to be detected and after heating. Based on the multiple frames of infrared images, the automatic encoding model is trained to obtain a target automatic encoding model, which can determine the degree of damage to the pixels in any infrared image. According to the multiple frames of infrared images and the target automatic encoding model, a characteristic image of the insulating equipment is constructed. The damage of the insulating equipment is detected according to the characteristic image, which has the advantage of intelligence. The detection method based on infrared images in the present application has the characteristic of being able to detect online at a long distance, so the need for power outage detection of power equipment is avoided, thereby avoiding affecting the normal use of the power equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0068] Figure 1 is a schematic diagram of a system architecture according to an exemplary embodiment;

[0069] Figure 2 is a flow chart of a method for detecting insulation equipment according to an exemplary embodiment;

[0070] Figure 3 is a flow chart of a method for detecting insulation equipment according to another exemplary embodiment;

[0071] Figure 4 is a schematic diagram showing a detection result of an insulation device according to an exemplary embodiment;

[0072] Figure 5 is a schematic diagram showing comparison of characteristic images according to an exemplary embodiment;

[0073] Figure 6 is a schematic diagram showing comparison of binary images according to an exemplary embodiment;

[0074] Figure 7 is a structural schematic diagram of a device for detecting insulation equipment according to an exemplary embodiment;

[0075] Figure 8 The diagram is a schematic structural diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0076] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below in conjunction with the accompanying drawings.

[0077] It should be understood that the "multiple" mentioned in this application refers to two or more. In the description of this application, unless otherwise specified, " / " means or, for example, A / B can mean A or B; "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in order to facilitate the clear description of the technical solution of this application, the words "first" and "second" are used to distinguish between the same items or similar items with basically the same functions and effects. Those skilled in the art can understand that the words "first" and "second" do not limit the quantity and execution order, and the words "first" and "second" do not limit them to be different.

[0078] Before introducing the method provided in the embodiment of the present application in detail, the system architecture involved in the embodiment of the present application is first introduced. Figure 1 , Figure 1The system architecture is a schematic diagram according to an exemplary embodiment, which includes an industrial computer 110, a thermal excitation source 120, a motor 130, an infrared thermal imager 140, and an electronic device 150. In one example, the electronic device 150 and the infrared thermal imager 140 establish a communication connection via wired or wireless communication.

[0079] As an example of the present application, the industrial computer 110 is used to generate a control signal to control the thermal excitation source 120 to generate thermal radiation, such as generating infrared light above 2.5 microns. The thermal radiation is focused on the insulating device to be detected under the action of a parabolic mirror (for example, provided at the light and heat outlet of the thermal excitation source) to form a strip heating of the insulating device. In addition, during the heating process, the industrial computer 110 controls the motor 130 to rotate so that the motor 130 drives the thermal excitation source 120 to rotate and heat, thereby realizing surface heating of the insulating device. After the thermal excitation source 120 heats for a preset time, it stops heating, that is, the heating ends, so that the thermal excitation source 120 generates a period of constant thermal excitation on the insulating device. Afterwards, the industrial computer 110 controls the infrared thermal imager 140 to collect and obtain the radiation data of the insulating device during the cooling process according to the preset acquisition frequency, that is, to collect the surface temperature of the insulating device during the temperature drop period, and obtain multiple frames of infrared images, so that the electronic device 150 can obtain multiple frames of infrared images from the infrared thermal imager 140, and judge the damage of the insulating device based on the acquired multiple frames of infrared images.

[0080] Among them, the preset duration can be set by the user according to actual needs, or can also be set by default by the industrial computer, and the embodiments of the present application are not limited to this.

[0081] Among them, the preset acquisition frequency can be set by the user according to actual needs, or can also be set by default by the industrial computer, and the embodiments of the present application are not limited to this.

[0082] In one embodiment, thermal activation source 120 may be a device such as an infrared halogen lamp.

[0083] In one embodiment, the electronic device 150 may include but is not limited to terminal devices such as a tablet computer, a laptop computer, and a desktop computer, which is not limited in this embodiment of the present application.

[0084] Based on the system architecture provided above, the detection process of the insulation equipment provided by the embodiment of the present application is described in detail below. Figure 2 , Figure 2 1 is a flow chart of a method for detecting an insulating device according to an exemplary embodiment. As an example but not a limitation, the method is performed by the electronic device 150, and the method may include the following steps:

[0085] Step 201: Acquire multiple frames of infrared images, where the multiple frames of infrared images are acquired by capturing images of the insulation device to be inspected and after being heated.

[0086] As mentioned above, in order to detect the damage of the insulation equipment, the thermal excitation source can be controlled by the industrial computer to heat the insulation equipment to be detected. After the heating is completed, during the cooling process of the insulation equipment, the infrared thermal imager collects data to obtain an infrared image of the insulation equipment. The electronic device obtains multiple frames of infrared images collected from the infrared thermal imager.

[0087] In one embodiment, the acquired multiple frames of infrared images include infrared images after the k-th frame, where k is an integer greater than 1.

[0088] Specifically, since there is still a certain amount of residual heat in the thermal excitation source within a period of time just after the heating ends (for example, within a period of about 3 seconds), the infrared image captured by the infrared thermal imager may be inaccurate due to the influence of the residual heat of the thermal excitation source during this period, that is, the infrared image captured during this period fails to accurately reflect the actual temperature of the insulation equipment. Therefore, the infrared images during this period can be filtered out, and the infrared images captured during the period after the thermal excitation source cools down can be retained. The electronic device performs subsequent detection operations based on the retained multiple frames of infrared images, which can improve the accuracy of the detection.

[0089] In one embodiment, the image acquisition time of the acquired multiple frames of infrared images is continuous, and the image acquisition time interval between each two adjacent frames of infrared images is the same. For example, the first frame of the multiple frames of infrared images is acquired at the 4th second, the second frame is acquired at the 5th second, the third frame is acquired at the 6th second, and so on.

[0090] Step 202: Based on multiple frames of infrared images, the automatic encoding model is trained to obtain a target automatic encoding model, and the target automatic encoding model can determine the damage degree of the pixel points in any infrared image.

[0091] A simple autoencoder (AE) usually includes an input layer, a hidden layer, and an output layer. The output layer and the input layer can have the same number of nodes, and the hidden layer has a smaller number of nodes. The high-dimensional data input is encoded between the input layer and the hidden layer to reduce the dimensionality of the high-dimensional data, that is, to reduce the high-dimensional data to low-dimensional data. This operation process can be called an encoder here; the low-dimensional data after dimensionality reduction is decoded between the hidden layer and the output layer to restore the low-dimensional data to high-dimensional data. This operation process can be called a decoder here. Through AE, we hope to obtain a lossy dimensionality reduction representation of high-dimensional data, requiring that the high-dimensional data restored by the decoder during training is as similar as possible to the original input data.

[0092] In one embodiment, multiple simple AEs can be cascaded, for example, a cascade of AEs with three hidden layers can be used to obtain a deep autoencoder (DAE) as an autoencoding model. The meaning of cascading is to use AE model 2 to further reduce the dimension of the hidden layer data of AE model 1, wherein the hidden layer of AE model 1 is used as the input layer of AE model 2; AE model 3 further reduces the dimension of the hidden layer data of AE model 2, wherein the hidden layer of AE model 2 is used as the input layer of AE model 3; if there are more hidden layers, the same applies. The number of nodes in each hidden layer is given in advance, and training is performed using a layer-by-layer greedy training method.

[0093] That is to say, before applying the automatic encoding model, the automatic encoding model is first trained using multiple frames of infrared images acquired by the electronic device, such as Figure 3 As shown. In this way, the target automatic encoding model obtained after training can grasp the temperature change law of pixels representing normal areas, so that the target automatic encoding model can determine the degree of damage of pixels in any infrared image. Since the number of pixels representing normal areas is much greater than the number of pixels representing abnormal areas in actual detection scenarios, randomly selected pixels can be equivalent to pixels representing normal areas.

[0094] In one embodiment, the specific implementation of training the automatic encoding model based on multiple infrared image frames may include: performing median filtering on each infrared image frame in the multiple infrared image frames. Selecting a detection area in each infrared image frame after the median filtering process to obtain multiple local area images, where the detection area is the area where the insulation device is located. Based on the multiple local area images, the automatic encoding model is trained.

[0095] In order to suppress the influence of salt and pepper noise outliers, the electronic device performs median filtering on multiple infrared images frame by frame. Figure 3 As shown, the detection area containing insulating equipment is retained, such as the porcelain bushing detection area, and the rest of the background area is deleted. Subsequently, vectorization and sampling can be performed to obtain a data set for neural network input.

[0096] In one embodiment, the detection area in each infrared image frame may be extracted by a target detection model, etc., to obtain multiple detection areas. The target detection model may be a pre-trained model that can be used to detect the detection area in any infrared image.

[0097] In one example, based on multiple local area images, the specific implementation of training the automatic encoding model may include: the pixel values ​​of each pixel at the same position in the multiple local area images are formed into a t-dimensional vector in the order of image acquisition time. The number of t multiple local area images can obtain m such vectors, where m is the number of pixels included in the detection area. K vectors are randomly selected from the m vectors, and k can be any value between m / 20 and m / 10. These vectors are concatenated to form a k*t matrix. The matrix is ​​normalized as a whole, and the automatic encoding model is trained based on the normalized k vectors.

[0098] For details, please continue to refer to Figure 3 , the electronic device constructs a t-dimensional vector by using the pixel values ​​of the pixel points located at the first row and first column of multiple local area images in the order of image acquisition time. The pixel values ​​of the pixel points located at the first row and second column of multiple local area images are constructed into a t-dimensional vector in the order of image acquisition time. The pixel values ​​of the pixel points located at the first row and third column of multiple local area images are constructed into a t-dimensional vector in the order of image acquisition time. By analogy, m t-dimensional vectors can be constructed. For example, assuming that the size of each detection area is 400*300, 120,000 t-dimensional vectors can be constructed, each of which can be used to indicate the temperature change process of the pixel point at the same position.

[0099] In one embodiment, since the number of determined vectors is large, in order to ensure effective training and improve training efficiency, k vectors may be randomly selected from the obtained m vectors, and the selected k vectors may be normalized as a whole, and then the normalized k vectors may be used as training data for training the automatic encoding model. The training data includes normal data and abnormal data (damage area data), wherein the abnormal data representing the damaged area has a larger reconstruction error, so that the trained automatic encoding model can distinguish between normal data and abnormal data.

[0100] During training, the loss function of the auto-encoding model is set. For example, the loss function is as shown in formula (1):

[0101]

[0102] Among them, L is the MSE (Mean-square error) loss, x i is the i-th input data, g θ represents the encoder, h θ represents the decoder, ω ijis the weight matrix, which is the weight representation from the i-th node to the j-th node, λ is the preset empirical value, and n is the number of input data. The right side of the formula is the weight regularization term, which can deal with the overfitting problem and improve robustness.

[0103] After that, the normalized k vectors are output in batches to the auto-encoding model for training. The purpose of the training is to learn the low-dimensional features of the original high-dimensional time series data to learn the model parameters that minimize the loss function. When the training end condition is met, the training ends, and the target auto-encoding model is obtained.

[0104] Among them, the training end condition can be set according to actual needs. For example, the training end condition can refer to the number of training times reaching a preset number threshold, etc., which is not limited in this embodiment of the present application.

[0105] Step 203: construct a feature image of the insulation device according to the multiple frames of infrared images and the target automatic encoding model.

[0106] In one example, the specific implementation of step 203 may include: the pixel values ​​of each pixel point at the same position in the multiple local area images are combined into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixels included in the detection area. The m vectors are input into the trained target auto-encoding model, that is, the target auto-encoding model is applied to all m vectors, and the MSE of the input and output is used as the eigenvalue, and a total of m eigenvalues ​​are output, and each of the m eigenvalues ​​is used to indicate the degree of damage of the corresponding pixel point. Based on the m eigenvalues, a characteristic image of the insulation equipment is constructed.

[0107] As an example of the present application, the electronic device applies the target auto-encoding model to each pixel, and uses the MSE loss of each pixel as the feature value to construct a feature image. Specifically, since the target auto-encoding model can distinguish between normal data and abnormal data, the vector of each pixel can be input into the target auto-encoding model, and the target auto-encoding model outputs a vector of the same dimension after dimension reduction and dimension recovery. According to the MSE loss between the output vector and the input vector, the degree of damage of each pixel can be determined. For example, if a vector is input into the target auto-encoding model, if it is determined that the MSE loss corresponding to the vector is large, it means that the output vector and the vector are very different, that is, the target auto-encoding model cannot recover similar data after the dimension reduction of the vector. In this case, it means that the area where the pixel corresponding to the vector is located has a large degree of damage. The electronic device uses the MSE loss of each pixel as the feature value of each pixel. In this way, after determining the feature value of each pixel among all pixels, the corresponding feature image can be constructed.

[0108] Step 204: Detect damage to the insulation device based on the feature image.

[0109] As an example of the present application, the specific implementation of step 204 may include: performing binary segmentation processing on the feature image to obtain a binary image, determining all connected regions in the binary image, and detecting the damage of the insulation equipment according to the area of ​​each connected region in all connected regions.

[0110] The damaged area has the following characteristics: the grayscale value of the damaged area is significantly higher than that of the surrounding area; the damaged area has a certain area and does not appear in the form of isolated points; most of the feature images are normal areas. Figure 3 As shown, after obtaining the feature image, the electronic device determines the binary image based on the feature image. Specifically, the electronic device can use an adaptive threshold method to perform binary segmentation processing on the feature image, that is, for each pixel in the feature image, the neighborhood mean of each pixel is added or subtracted by a preset offset, and the obtained result is used as the judgment threshold of each pixel to determine the binary image.

[0111] Afterwards, all connected components in the binary image are extracted to determine all connected regions in the binary image. In this way, the electronic device can detect the damage of the insulating device according to the area of ​​each connected region in all connected regions.

[0112] As an example of the present application, the specific implementation of detecting the damage condition of the insulating equipment based on the area of ​​each connected area among all connected areas may include: if there is a connected area among all connected areas whose area is greater than or equal to an area threshold, it is determined that the insulating equipment is seriously damaged.

[0113] Among them, the area threshold can be set by the user according to actual needs, or can also be set by default by the electronic device, and the embodiments of the present application are not limited to this.

[0114] If the area of ​​a connected region is greater than the area threshold, it means that the location corresponding to the connected region in the insulating device has a large area of ​​damage, and it can be determined that the insulating device is seriously damaged. On the contrary, if the area of ​​each of the connected regions determined is less than the area threshold, it means that the location corresponding to each connected region in the insulating device has only a small area of ​​damage, which can be ignored, and it can be determined that the insulating device is not seriously damaged.

[0115] In one embodiment, when it is determined that the insulating equipment is seriously damaged, an early warning can be issued, for example, by sending a message, displaying a message, ringing a bell, etc., to remind the user that there is a problem with the insulating equipment currently being detected, so that relevant personnel can take corresponding measures in time.

[0116] It is worth mentioning that in some other embodiments, infrared images are usually obtained, and then the damage of the insulating equipment is detected based on the infrared images, for example, based on the infrared images, the detection is performed by a method of manual selection of original thermal images, frequency domain transformation method, and principal component analysis method. However, the false detection rate of these methods is relatively large. Compared with these methods, the present application uses an automatic encoding model for detection, which can eliminate background interference to a great extent and significantly improve the feature image contrast and signal-to-noise ratio.

[0117] Next, the advantages of the method provided by the embodiment of the present application are explained by taking the composite insulating flat plate as an example. Figure 4 The figure shows a double-layer structure, with the upper layer being high-temperature vulcanized silicone rubber and the lower layer being epoxy resin, and the upper and lower layers being bonded by a coupling agent. There are holes with different diameters (D) and depths (H) at the interface. The automatic encoding model has the smallest reconstruction error for defect-free pixels, and the expected effect is good. Figure 5 From left to right in the figure are the feature images based on infrared images, the original thermal image manual selection method, the frequency domain transformation amplitude method, the frequency domain transformation phase method, the principal component analysis method, and the automatic encoding model method. Figure 6 From left to right in the figure are the binarized images based on infrared images, the original thermal image manual selection method, the frequency domain transformation amplitude method, the frequency domain transformation phase method, the principal component analysis method, and the automatic encoding model method. The results show that the automatic encoding model greatly eliminates background interference compared with other methods and significantly improves the feature image contrast and signal-to-noise ratio; in the binary image, the false detection rate of the automatic encoding model detection is lower than that of other methods. Combined with its unsupervised and intelligent feature selection characteristics, the intelligent detection scheme based on the automatic encoding model can improve the accuracy of transient infrared detection, provide important reference for detection personnel, and has important engineering significance.

[0118] In an embodiment of the present application, multiple frames of infrared images are obtained by collecting images of the insulating equipment to be detected and heated. Based on the multiple frames of infrared images, the automatic encoding model is trained to obtain a target automatic encoding model, which can determine the degree of damage to the pixels in any infrared image. According to the multiple frames of infrared images and the target automatic encoding model, a characteristic image of the insulating equipment is constructed. The damage of the insulating equipment is detected according to the characteristic image. The detection method based on infrared images in the present application has the characteristics of online long-distance detection, so it avoids the need for power outage detection of power equipment, thereby avoiding affecting the normal use of power equipment. This method can be applied to the detection of insulating equipment in the power industry, which can effectively prevent defects such as insulators from further expanding and causing accidents, and is conducive to the safe and stable operation of the power grid. In addition, transient thermal imaging nondestructive testing also has the advantages of large single detection area and simple operation.

[0119] It should be understood that the serial numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0120] Based on the above methods, please refer to Figure 7 , Figure 7 This is a schematic diagram of a structure of a device for detecting insulation equipment according to an exemplary embodiment. The device for detecting insulation equipment can be implemented as part or all of an electronic device by software, hardware, or a combination of both. The device may include:

[0121] An acquisition module 710 is used to acquire multiple frames of infrared images, where the multiple frames of infrared images are acquired by collecting images of the insulation device to be inspected and after being heated;

[0122] A training module 720, configured to train the auto-encoding model based on the multiple frames of infrared images to obtain a target auto-encoding model, wherein the target auto-encoding model can determine the degree of damage of a pixel in any infrared image;

[0123] A construction module 730, configured to construct a feature image of the insulation device according to the multiple frames of infrared images and the target automatic encoding model;

[0124] The detection module 740 is used to detect the damage of the insulation device according to the characteristic image.

[0125] As an example of the present application, the training module 720 is used to:

[0126] Performing median filtering on each infrared image frame in the multiple infrared image frames;

[0127] Selecting a detection area in each frame of the infrared image after median filtering to obtain a plurality of local area images, wherein the detection area is the area where the insulation equipment is located;

[0128] Based on the multiple local area images, training the auto-encoding model;

[0129] The construction module 730 constructs the characteristic image of the insulation device according to the multiple frames of infrared images and the target automatic encoding model, including:

[0130] The construction module 730 constructs a feature image of the insulation device according to the multiple local area images and the target automatic encoding model.

[0131] As an example of the present application, the training module 720 is used to:

[0132] Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area;

[0133] Select k vectors from the m vectors, where k is an integer greater than 1 and less than m;

[0134] Combining the k vectors into a matrix;

[0135] Normalizing the matrix;

[0136] The automatic encoding model is trained based on the k normalized vectors.

[0137] As an example of the present application, the building module 720 is used to:

[0138] Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area;

[0139] Inputting the m vectors into the target auto-encoding model, and outputting m eigenvalues, each of the m eigenvalues ​​being used to indicate the degree of damage of a corresponding pixel point;

[0140] A characteristic image of the insulation device is constructed based on the m characteristic values.

[0141] As an example of the present application, the detection module 740 is used to:

[0142] Performing a binary segmentation process on the feature image to obtain a binary image;

[0143] Determine all connected regions in the binary image;

[0144] The damage condition of the insulation device is detected according to the area of ​​each connected region among all the connected regions.

[0145] As an example of the present application, the detection module 740 is used to:

[0146] If there is a connected area whose area is greater than or equal to the area threshold among all the connected areas, it is determined that the insulation device is seriously damaged.

[0147] As an example of the present application, the multiple frames of infrared images include infrared images after the kth frame, where k is an integer greater than 1.

[0148] In an embodiment of the present application, multiple frames of infrared images are acquired, and the multiple frames of infrared images are acquired by capturing images of the insulating equipment to be detected and after being heated. Based on the multiple frames of infrared images, the automatic encoding model is trained to obtain a target automatic encoding model, and the target automatic encoding model can determine the degree of damage to the pixels in any infrared image. According to the multiple frames of infrared images and the target automatic encoding model, a characteristic image of the insulating equipment is constructed. The damage of the insulating equipment is detected based on the characteristic image. The detection method based on infrared images of the present application has the characteristic of being able to detect online at a long distance, so the need for power outage detection of power equipment is avoided, thereby avoiding affecting the normal use of the power equipment.

[0149] Figure 8 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 8 As shown, the electronic device 8 of this embodiment includes: at least one processor 80 ( Figure 8 Only one is shown), a memory 81, and a computer program 82 stored in the memory 81 and executable on the at least one processor 80, wherein the processor 80 implements the steps of any of the above-mentioned method embodiments when executing the computer program 82.

[0150] The electronic device 8 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The electronic device may include, but is not limited to, a processor 80 and a memory 81. Those skilled in the art will appreciate that Figure 8 It is only an example of the electronic device 8 and does not constitute a limitation on the electronic device 8. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0151] The processor 80 may be a CPU (Central Processing Unit), or other general-purpose processors, DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0152] In some embodiments, the memory 81 may be an internal storage unit of the electronic device 8, such as a hard disk or memory of the electronic device 8. In other embodiments, the memory 81 may also be an external storage device of the electronic device 8, such as a plug-in hard disk, SMC (Smart Media Card), SD (Secure Digital) card, Flash Card, etc. equipped on the electronic device 8. Further, the memory 81 may also include both an internal storage unit and an external storage device of the electronic device 8. The memory 81 is used to store an operating system, an application program, a boot loader, data, and other programs, such as the program code of the computer program. The memory 81 may also be used to temporarily store data that has been output or is to be output.

[0153] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0154] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0155] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person 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 spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for detecting insulation equipment, characterized in that: The method comprises: Acquire multiple frames of infrared images, wherein the multiple frames of infrared images are acquired by collecting images of the insulation device to be inspected and after being heated; Performing median filtering on each infrared image frame in the multiple infrared image frames; Selecting a detection area in each frame of the infrared image after median filtering to obtain a plurality of local area images, wherein the detection area is the area where the insulation equipment is located; Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area; Select k vectors from the m vectors, where k is an integer greater than 1 and less than m; Combining the k vectors into a matrix; Normalizing the matrix; Based on the k normalized vectors, an automatic encoding model is trained to obtain a target automatic encoding model, wherein the target automatic encoding model can determine the degree of damage of a pixel point in any infrared image; Combining the pixel values ​​of each pixel point at the same position in the multiple local area images into vectors in the order of image acquisition time to obtain m vectors, where m is the number of pixel points included in the detection area; Inputting the m vectors into the target auto-encoding model, and outputting m eigenvalues, each of the m eigenvalues ​​being used to indicate the degree of damage of a corresponding pixel point; Based on the m characteristic values, construct a characteristic image of the insulation device; The damage condition of the insulating device is detected according to the characteristic image.

2. The method according to claim 1, characterized in that Detecting the damage of the insulation device according to the characteristic image includes: Performing a binary segmentation process on the feature image to obtain a binary image; Determine all connected regions in the binary image; The damage condition of the insulation device is detected according to the area of ​​each connected region among all the connected regions.

3. The method according to claim 2, characterized in that Detecting the damage of the insulating device according to the area of ​​each connected region among all the connected regions includes: If there is a connected area whose area is greater than or equal to the area threshold among all the connected areas, it is determined that the insulation device is seriously damaged.

4. The method according to claim 1, characterized in that The multiple frames of infrared images include infrared images after the kth frame, where k is an integer greater than 1.

5. A detection device for insulating equipment, characterized in that: The device comprises: An acquisition module, used for acquiring multiple frames of infrared images, wherein the multiple frames of infrared images are acquired by acquiring images of the insulation device to be inspected and after being heated; A training module is used to perform median filtering on each infrared image in the multiple infrared images; select a detection area in each infrared image after the median filtering to obtain multiple local area images, where the detection area is the area where the insulation equipment is located; form vectors with pixel values ​​of each pixel point at the same position in the multiple local area images in the order of image acquisition time to obtain m vectors, where m is the number of pixels included in the detection area; select k vectors from the m vectors, where k is an integer greater than 1 and less than m; form a matrix with the k vectors; normalize the matrix; train an automatic encoding model based on the k vectors after the normalization to obtain a target automatic encoding model, where the target automatic encoding model can determine the degree of damage of the pixel points in any infrared image; A construction module, used to form a vector by combining the pixel values ​​of each pixel point at the same position in the multiple local area images in the order of image acquisition time, to obtain m vectors, wherein m is the number of pixel points included in the detection area; input the m vectors into the target automatic encoding model, and output m eigenvalues, wherein each eigenvalue of the m eigenvalues ​​is used to indicate the degree of damage of the corresponding pixel point; and construct a characteristic image of the insulation device based on the m eigenvalues; A detection module is used to detect damage to the insulation device according to the characteristic image.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images