A power equipment condition monitoring system and method based on image analysis

Abstract

The present invention belongs to the field of image analysis technology, and discloses a power equipment condition monitoring system and method based on image analysis; including: pre-processing the aging images by collecting insulation material aging images and their environmental parameters and electrical characteristic data, and then simulating different lighting and environmental conditions through data enhancement technology, including color temperature adjustment, brightness change, directional light source change, contrast optimization, local contrast enhancement, and noise injection, etc. These enhancement forms are selected by random numbering to avoid model overfitting and improve generalization ability, and use deep learning models, combined with processed secondary images and related data training, to predict the degree of aging in real time, and trigger an early warning when the preset threshold is reached to ensure the safe operation of the equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and more specifically, to a system and method for monitoring the condition of electric power equipment based on image analysis. Background Art

[0002] Power equipment refers to all mechanical devices and electronic components used for power generation, transmission, transformation, distribution and consumption. Insulation materials are key components of power equipment. Their main function is to isolate live conductors and prevent current leakage and short circuits. Over time, insulation materials will gradually age, which directly affects the performance and safety of power equipment.

[0003] The Chinese patent with the authorization announcement number CN116309603A discloses a method and system for visual detection of power equipment faults; by capturing the image at the initial moment and continuously shooting the real-time monitoring image after a fixed time interval, the dust accumulation on the insulator in the switch cabinet is observed over time. By comparing the initial image with the real-time monitoring image, it is evaluated whether the short-circuit risk level of the switch cabinet has reached a predetermined threshold, and the corresponding warning information is generated.

[0004] The Chinese patent with invention publication number CN117876781A discloses a method, system, device and medium for visualizing the aging state of an insulator surface. A convolutional neural network mathematical model is trained by preprocessing hyperspectral images and the aging level corresponding to each hyperspectral image to obtain a classification model, and an insulator surface aging state diagram is generated according to the aging level of each pixel. The problem that insulator aging is difficult to detect due to microscopic changes is solved.

[0005] The above patents all use image analysis, but fail to take into account that the image quality of a photo is closely related to the ambient light. When the ambient light changes, the color and clarity of the photo change significantly. A unified classification standard is difficult to use in practical applications, and the warning capability is not accurate enough.

[0006] In view of this, the present invention proposes a power equipment condition monitoring system and method based on image analysis to solve the above problems. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a system and method for monitoring the condition of electric power equipment based on image analysis.

[0008] To achieve the above object, the present invention provides the following technical solution: a method for monitoring the condition of electric power equipment based on image analysis, the method comprising:

[0009] Collect insulation material aging images, and collect environmental parameters and material electrical characteristic data corresponding to the insulation material aging images; environmental parameters include material temperature, temperature difference, humidity, altitude, ultraviolet intensity, and air quality index evaluation value; the temperature measured at the edge of a circle with a radius of 10 meters and a distance from the location of the insulation material as the center is marked as the environmental impact temperature, and the temperature difference is the difference between the environmental impact temperature and the material temperature; the air quality index evaluation value is obtained by the weighted sum of the air quality index; the material electrical characteristic data include surface resistance and partial discharge amplitude;

[0010] Calculate external parameter evaluation values ​​based on altitude, UV intensity and air quality index evaluation values;

[0011] Preprocessing insulation material aging images;

[0012] Mark the preprocessed insulation material aging image as a primary image; perform data enhancement based on the primary image; data enhancement includes: illumination transformation, contrast adjustment and noise injection; illumination transformation includes: color temperature adjustment, brightness change and directional light source change; contrast adjustment includes contrast optimization and local contrast enhancement; noise injection includes Gaussian noise, salt and pepper noise and Poisson noise;

[0013] Set corresponding numbers for the eight forms of data enhancement, including color temperature adjustment, brightness change, directional light source change, contrast optimization, local contrast enhancement, Gaussian noise, salt and pepper noise, and Poisson noise; set enhancement selection parameters and enhancement quantity parameters; before each primary image is input into the machine learning model, obtain the enhancement quantity parameter, and the enhancement quantity parameter randomly generates one of the integers from 1 to 8; it represents the number of data enhancement forms to be performed; at the same time, the enhancement selection parameter randomly generates one of the integers from 1 to 8, representing the number of data enhancement forms to be performed; process the primary image according to the data enhancement form corresponding to the number; at the same time, determine whether the requirements of the enhancement quantity parameter are met, and if not, regenerate the enhancement selection parameter; select the next data enhancement form to be executed; when the data enhancement form is repeated, regenerate it until the number of data enhancement forms meets the requirements of the enhancement quantity parameter, and mark the primary image at this time as a secondary image;

[0014] Based on the secondary images and the corresponding material temperature, temperature difference, humidity, external parameter evaluation values ​​and material electrical characteristics data, a machine learning model is trained to predict the material aging degree value for early warning processing.

[0015] Preferably, UVA and UVB intensities are collected; the ultraviolet intensity is obtained by adding the UVA intensity multiplied by 0.8 and the UVB intensity multiplied by 0.2;

[0016] Calculation process of external parameter evaluation value:

[0017] Where: WJ is the external parameter evaluation value, H is the altitude, ZW is the ultraviolet intensity, and KQ is the air quality index.

[0018] Preferably, the pretreatment comprises:

[0019] Denoising: Use non-local means or BM3D algorithm to reduce the noise of insulation material aging images;

[0020] Contrast enhancement: Adaptive histogram equalization is used to improve the contrast of insulation material aging images;

[0021] Normalization: Normalize the pixel values ​​of the insulation material aging image to the range of [0,1].

[0022] Preferably, the color temperature adjustment includes:

[0023] Convert the primary image to CIE XYZ color space;

[0024] Apply Bradford transformation matrix to simulate the adaptation of human eyes to different light sources;

[0025] Calculate the transformation matrix according to the color temperature: According to the color temperature of the image, calculate a Bradford transformation matrix corresponding to a specific color temperature; the specific color temperature is set by material science experts and electrical experts, ranging from 4000K to 6000K;

[0026] Convert the transformed XYZ value back to RGB color space; Convert the XYZ value adjusted by color temperature back to RGB color space through the inverse transformation matrix;

[0027] The brightness change includes:

[0028] 1. Color space conversion: convert the image to HSV or LAB color space;

[0029] 2. Brightness channel adjustment: linear or nonlinear transformation of V channel or L channel;

[0030] 3. Gamma correction: Use gamma correction to simulate the nonlinear perception of brightness changes by the human eye;

[0031] 4. Color space conversion: convert the modified image back to the original color space;

[0032] The process of directional light source change includes: creating a gradient map with the same size as the first-level image; using trigonometric functions to generate the lighting pattern; alpha blending or multiplication of the gradient map with the original image: combining the lighting information in the gradient map with the pixel values ​​of the original image to simulate the lighting effect.

[0033] Preferably, the contrast optimization process includes: a. using histogram analysis to determine the contrast range of the image: analyzing the histogram of the image to understand the distribution of bright and dark pixels in order to adjust the contrast; b. applying gamma correction; c. using linear contrast stretching to enhance the contrast; d. using the S-curve of the sigmoid function to perform nonlinear contrast adjustment.

[0034] Preferably, the secondary image and the corresponding material temperature, temperature difference, humidity, external parameter evaluation value and material electrical characteristic data are converted into feature vectors, the material aging degree value is used as the label corresponding to the feature vector, the material aging degree value is an integer of 1-5, each set of feature vectors and the label corresponding to each set of feature vectors are constructed as a sample, and multiple samples are collected to construct a data set; the data set is divided into a training set, a validation set and a test set, wherein the training set accounts for 60% of the data set, and the validation set and the test set each account for 20% of the data set;

[0035] The training set is used as the input of the machine learning model, and the machine learning model uses the real-time material aging degree value as the output; the historical material aging degree value corresponding to a set of real-time feature vectors is used as the prediction target, and minimizing the machine learning model loss function value is used as the training target; when the machine learning model loss function value is less than or equal to the preset target loss value, the training is stopped;

[0036] The machine learning model loss function is the mean square error; the mean square error is calculated by Minimize to train the model; in the loss function, MSE is the loss function value, i is the feature vector group number; u is the number of feature vector groups; y i is the label corresponding to the i-th group of feature vectors, The material aging degree value predicted by the i-th group of feature vectors;

[0037] The machine learning model is a deep neural network model;

[0038] Early warning measures include arranging staff to replace insulation materials.

[0039] A power equipment condition monitoring system based on image analysis, the system comprising:

[0040] A data collection module collects aging images of insulating materials, and at the same time collects environmental parameters and material electrical characteristic data corresponding to the aging images of insulating materials; environmental parameters include material temperature, temperature difference, humidity, altitude, ultraviolet intensity, and air quality index evaluation value; the temperature measured at the edge of a circle with a radius of 10 meters and a distance from the location of the insulating material as the center is marked as the environmental impact temperature, and the temperature difference is the difference between the environmental impact temperature and the material temperature; the air quality index evaluation value is obtained by the weighted sum of the air quality index; the material electrical characteristic data include surface resistance and partial discharge amplitude;

[0041] A calculation module calculates the external parameter evaluation value based on the altitude, ultraviolet intensity and air quality index evaluation value;

[0042] A preprocessing module is used to preprocess the insulation material aging image;

[0043] The data enhancement module marks the preprocessed insulation material aging image as a primary image; performs data enhancement based on the primary image; data enhancement includes: illumination transformation, contrast adjustment and noise injection; illumination transformation includes: color temperature adjustment, brightness change and directional light source change; contrast adjustment includes contrast optimization and local contrast enhancement; noise injection includes Gaussian noise, salt and pepper noise and Poisson noise;

[0044] The tuning module sets corresponding numbers for the eight forms of data enhancement, including color temperature adjustment, brightness change, directional light source change, contrast optimization, local contrast enhancement, Gaussian noise, salt and pepper noise, and Poisson noise; sets enhancement selection parameters and enhancement quantity parameters; obtains the enhancement quantity parameter before each primary image is input into the machine learning model, and the enhancement quantity parameter randomly generates one of the integers from 1 to 8; represents the number of data enhancement forms to be performed; at the same time, the enhancement selection parameter randomly generates one of the integers from 1 to 8, representing the number of data enhancement forms to be performed; processes the primary image according to the data enhancement form corresponding to the number; at the same time, determines whether the requirements of the enhancement quantity parameter are met, and if not, regenerates the enhancement selection parameter; selects the next data enhancement form to be executed; when the data enhancement form is repeated, regenerates it until the number of data enhancement forms meets the requirements of the enhancement quantity parameter, and marks the primary image at this time as a secondary image;

[0045] The model training module trains a machine learning model to predict the material aging degree based on the secondary image and the corresponding material temperature, temperature difference, humidity, external parameter evaluation value and material electrical characteristic data for early warning processing.

[0046] An electronic device comprises: a processor and a memory, wherein the memory stores a computer program that can be called by the processor; the processor executes the above-mentioned method for monitoring the condition of electric power equipment based on image analysis by calling the computer program stored in the memory.

[0047] A computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for monitoring the condition of electric power equipment based on image analysis.

[0048] The technical effects and advantages of the power equipment condition monitoring system and method based on image analysis of the present invention are as follows: by real-time monitoring of the aging images of insulating materials, and combining environmental parameters and material electrical characteristics data, the aging degree is predicted using a machine learning model; the aging images are collected and preprocessed, including denoising, contrast enhancement and standardization; data enhancement is performed, different lighting and environmental conditions are simulated, and deep neural network model training is used to predict the aging degree in real time. When the predicted result reaches the aging threshold, an early warning is triggered and corresponding processing is performed. The advantage of this solution is that it comprehensively considers multi-dimensional environmental factors and material properties, and improves the accuracy and reliability of the prediction. At the same time, the generalization ability of the model for different environmental conditions is enhanced, and the risk of overfitting is reduced. At the same time, the real-time monitoring and early warning mechanism ensures the safety of insulating materials and the stable operation of equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic diagram of a method for monitoring the condition of electric power equipment based on image analysis according to the present invention;

[0050] Figure 2 A schematic diagram of a power equipment condition monitoring system based on image analysis according to the present invention;

[0051] Figure 3 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application;

[0052] Figure 4 It is a schematic diagram of the structure of a computer-readable storage medium provided by an embodiment of the present application. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Example 1

[0055] See also Figure 1 As shown, the present embodiment provides a method for monitoring the condition of electric power equipment based on image analysis, the method comprising:

[0056] Collect insulation material aging images, and at the same time collect environmental parameters and material electrical characteristic data corresponding to the insulation material aging images; the environmental parameters and material electrical characteristic data can be used as additional input features of the machine learning model to improve the accuracy of the prediction.

[0057] The insulation material aging images are acquired by real-time monitoring using a camera installed next to the insulation material;

[0058] Environmental parameters include material temperature, temperature difference, humidity, altitude, UV intensity, and air quality index assessment;

[0059] The material temperature is the temperature at the location of the insulating material, which is obtained by a temperature sensor installed at the location of the insulating material;

[0060] The temperature measured at the edge of a circle with a radius of 10 meters and a distance from the insulating material as the center is marked as the environmental impact temperature, and the temperature difference is the difference between the environmental impact temperature and the material temperature;

[0061] It is understandable that the insulating material is located in a distribution box or other electrical integrated equipment. The temperature of the insulating material in the integrated equipment is often affected by the external temperature and is also related to its own loss. When the temperature of the insulating material inside the integrated equipment is too different from that of the outside world, the reliability of the insulating material is greatly reduced and requires precise monitoring. At the same time, when the temperature of the material itself is high, it may also be that the ambient temperature itself is high. In this case, the impact of the environment needs to be eliminated. It cannot be unilaterally stated that there is a problem with the insulating material.

[0062] The humidity is the humidity at the location of the insulating material, and the humidity is recorded using a humidity sensor.

[0063] The altitude is the height of the location of the insulating material, which is obtained by GPS positioning.

[0064] UV intensity: Use an ultraviolet radiometer to measure UVA and UVB intensity. The UV intensity is calculated by adding the UVA intensity multiplied by 0.8 and the UVB intensity multiplied by 0.2. The reason for this setting is that UVA has a longer wavelength and stronger penetrating power, and can penetrate clouds and glass. UVB has a shorter wavelength and weaker penetrating power, and most of it is absorbed by the ozone layer.

[0065] The air quality index is obtained from real-time data released by the Meteorological Bureau, including the concentrations of PM2.5, PM2.5 (fine particulate matter), PM10 (inhalable particulate matter), SO2, NO2, O3 and CO; the air quality index assessment value is obtained by the weighted sum of the air quality index; the weight of the weighted sum is set by material experts based on the impact of pollutants on the aging of insulating materials. The greater the impact, the greater the corresponding weight, and vice versa.

[0066] Calculate external parameter evaluation values ​​based on altitude, UV intensity and air quality index evaluation values In the formula: WJ is the external parameter evaluation value, H is the altitude, ZW is the ultraviolet intensity, and KQ is the air quality index; the higher the altitude, the thinner the air is, the closer the straight-line distance to the sun is, the higher the risk of insulation material aging, the higher the ultraviolet intensity, the higher the air quality index evaluation value, which means that the worse the air environment, the more likely the insulation material will age, and the higher the external parameter evaluation value.

[0067] Material electrical properties data include surface resistance and partial discharge amplitude;

[0068] The surface resistance is obtained by measuring the insulating material with a surface resistance tester; the partial discharge amplitude is obtained by real-time monitoring using high-frequency current monitoring (HFCT) technology;

[0069] High-frequency current transformers (HFCTs) work based on Faraday's law of electromagnetic induction. When partial discharge occurs, it rapidly generates current pulses that appear as high-frequency signals. HFCTs are able to detect these high-frequency current disturbances and convert them into voltage signals for further analysis and processing.

[0070] The partial discharge amplitude refers to the magnitude of the discharge current during the partial discharge process. It is an important parameter for evaluating the intensity of partial discharge activity and can reflect the condition of the insulation material and the operating status of the equipment.

[0071] It should be noted that material temperature is a key factor affecting the aging of insulating materials. High temperature can accelerate chemical reactions, resulting in a decrease in material performance. Real-time monitoring of material temperature can help predict its aging rate. The temperature difference can reflect the difference between the insulating material and the external environment, which is crucial for predicting the aging rate of the material under different environmental conditions. Humidity can affect the water absorption and electrical properties of the insulating material, so monitoring humidity can help assess the aging state of the material. Altitude affects air pressure and temperature, which in turn affect the aging of the insulating material. The aging risk of insulating materials in high altitude areas is higher. Ultraviolet rays can accelerate the aging of insulating materials, especially for certain materials such as polymers. Real-time monitoring of ultraviolet intensity can help assess the aging risk. The air quality index assessment value reflects the degree of environmental pollution, and the accumulation of pollutants can accelerate the aging of insulating materials. Surface resistance can reflect the insulating performance of insulating materials. Reduced surface resistance may mean the aging of insulating materials.

[0072] Preprocessing the insulation material aging image; the preprocessing includes:

[0073] Denoising: Use the non-local means (NLM) or BM3D algorithm to reduce the noise of the insulation material aging image.

[0074] Remove noise from images, improve image quality, and make subsequent image analysis and recognition tasks more accurate.

[0075] Contrast enhancement: Adaptive histogram equalization (CLAHE) is used to improve the contrast of insulation material aging images. Improving the contrast of the image makes the features in the image more prominent, which is convenient for subsequent image analysis and recognition.

[0076] Standardization: Normalize the pixel values ​​of the insulation material aging image to the range of [0,1]. Convert the pixel values ​​of the image to a specific range to facilitate model training and prediction.

[0077] Mark the preprocessed insulation material aging image as a primary image; perform data enhancement based on the primary image;

[0078] Data augmentation includes: illumination transformation, contrast adjustment, and noise injection;

[0079] Lighting changes include: color temperature adjustment, brightness change and directional light source change;

[0080] Color temperature adjustments include:

[0081] Convert the primary image to CIE XYZ color space;

[0082] The RGB color space is based on the three primary color channels of red, green, and blue, while the CIE XYZ color space is defined by the International Commission on Illumination, which is the mathematical basis of the RGB color space.

[0083] The Bradford transformation matrix is ​​used to simulate the adaptation of the human eye to different light sources; the Bradford transformation matrix is ​​used to process the converted XYZ values. The Bradford transformation matrix can simulate the adaptation of the human eye to different light sources, that is, the human eye will perceive color differently under different light sources. This process can compensate for this difference, making the image look more consistent under different light sources.

[0084] Calculate the transformation matrix based on the color temperature: Based on the color temperature of the image, calculate a Bradford transformation matrix corresponding to a specific color temperature. The specific color temperature is set by material science experts and electrical experts, ranging from 4000k to 6000k. Color temperature is a measure of the color of the light source, usually in Kelvin (k).

[0085] Convert the transformed XYZ values ​​back to RGB color space; convert the color temperature adjusted XYZ values ​​back to RGB color space through the inverse transformation matrix. The inverse transformation matrix is ​​the inverse operation of the forward transformation matrix, which can convert the transformed XYZ values ​​back to the original RGB values. This process allows the image to maintain consistent visual effects at different color temperatures, thereby improving the accuracy of image analysis and processing.

[0086] Through the color temperature adjustment process, the image can be made to look more consistent under different light sources, reducing the impact of light color temperature when taking pictures;

[0087] The brightness change includes:

[0088] 1. Color space conversion: Convert the image to HSV (hue, saturation, brightness) or LAB color space. These color spaces provide richer color representation than RGB color space.

[0089] 2. Brightness channel adjustment: perform linear or nonlinear transformation on the V channel (in HSV) or L channel (in LAB). The V channel or L channel contains the brightness information of the image. By adjusting these channels, the overall brightness of the image can be changed.

[0090] 3. Gamma correction: Use gamma correction (power law transformation) to simulate the nonlinear perception of brightness changes by the human eye. Gamma correction is a nonlinear transformation that can enhance the dark details of an image while maintaining the brightness of the bright parts.

[0091] 4. Color space conversion: Convert the modified image back to the original color space. This step ensures that the adjusted image can correctly represent the color and brightness information in the original color space.

[0092] The brightness change is to adjust the overall brightness of the image and simulate different lighting conditions to simulate the impact of different light intensities on visual perception.

[0093] Directional light changes produce shadow and highlight effects by simulating directional light sources.

[0094] The process of directional light source change includes:

[0095] Create a gradient map of the same size as the first-level image;

[0096] The gradient map is a two-dimensional array where each pixel value represents the light intensity at that location. The direction of the gradient represents the direction of the light source, while the strength of the gradient represents the intensity of the light.

[0097] Use trigonometric functions to generate more complex lighting patterns; Generate complex lighting patterns through trigonometric functions (such as sin and cos) to simulate the dynamic changes of light sources in the real world.

[0098] Trigonometric functions can be used to create periodic lighting patterns that simulate effects such as movement, rotation, or reflection of light sources. By adjusting the parameters of the trigonometric functions, the complexity and dynamics of the lighting pattern can be controlled.

[0099] Alpha blending or multiplication of the gradient map with the original image: combining the lighting information in the gradient map with the pixel values ​​of the original image to simulate lighting effects.

[0100] Alpha blending is an image synthesis technique that allows the pixel values ​​of two images to be mixed according to a certain transparency (alpha value). The multiplication operation multiplies the pixel values ​​of the gradient map with the pixel values ​​of the original image to simulate the lighting effect.

[0101] Adjust the blending strength to control the apparent intensity of the lighting effect: By adjusting the blending strength or alpha value, you can control the apparent intensity and strength of the lighting effect.

[0102] Mixed intensity or alpha value is an important parameter for controlling lighting effects. By adjusting these parameters, you can control the degree and intensity of the lighting effect, thereby achieving different lighting effects and visual effects.

[0103] Contrast adjustment includes contrast optimization and local contrast enhancement:

[0104] Contrast optimization improves image visibility by increasing or decreasing the difference between light and dark areas in the image.

[0105] The contrast optimization process includes: a. Using histogram analysis to determine the contrast range of the image: Analyze the image's histogram to understand the distribution of bright and dark pixels to facilitate contrast adjustment. b. Apply gamma correction. c. Use linear contrast stretching to enhance contrast. d. Use the S-curve of the sigmoid function for nonlinear contrast adjustment, which can maintain mid-tone details and avoid the destruction of details by contrast adjustment.

[0106] Local contrast enhancement is an adaptive contrast enhancement technique that enhances the local contrast of an image by local histogram equalization while limiting noise amplification.

[0107] Noise injection includes Gaussian noise, salt and pepper noise, and Poisson noise;

[0108] An exemplary noise injection process includes:

[0109] Gaussian noise: a. Generate a Gaussian distribution random array: Generate a Gaussian distribution random array of the same size as the image, with a mean of 0; b. Add the noise array to the original image to obtain a noisy image. c. Crop pixel values: Crop the result to ensure that the pixel values ​​are within the valid range.

[0110] Salt and pepper noise: a. Generate random array: Generate a random array of the same size as the image. b. Replace pixel values: Replace the values ​​in the random array with the maximum value (salt) or the minimum value (pepper) according to the preset threshold. c. Use mask to replace the original image: Use this array as a mask to selectively replace pixels in the original image.

[0111] Poisson noise: a. Treat pixel values ​​as Poisson distribution parameters: treat image pixel values ​​as the λ parameter of Poisson distribution. b. Sampling from Poisson distribution: for each pixel, sample a new value from the corresponding Poisson distribution. c. Scaling pixel values: scale the sampled result back to the brightness range of the original image.

[0112] Corresponding numbers are set for eight forms of data enhancement, including color temperature adjustment, brightness change, directional light source change, contrast optimization, local contrast enhancement, Gaussian noise, salt and pepper noise, and Poisson noise; illustratively, color temperature adjustment is set to 1, brightness change is set to 2, directional light source change is set to 3, and so on.

[0113] Set the enhancement selection parameter and enhancement quantity parameter; both the enhancement selection parameter and enhancement quantity parameter are integers from 1 to 8;

[0114] Before each level one image is input into the machine learning model, an enhancement quantity parameter is obtained. The enhancement quantity parameter randomly generates an integer from 1 to 8, which represents the number of data enhancement forms to be performed.

[0115] At the same time, the enhancement selection parameter randomly generates an integer from 1 to 8, representing the number of the data enhancement form to be performed;

[0116] The first-level image is processed according to the data enhancement form corresponding to the number; at the same time, it is determined whether the requirement of the enhancement quantity parameter is met, and if not, the enhancement selection parameter is regenerated; the next data enhancement form to be executed is selected; when the data enhancement form is repeated, it is regenerated until the number of data enhancement forms reaches the requirement of the enhancement quantity parameter, and the first-level image at this time is marked as a second-level image.

[0117] By introducing randomness, the model will not be overly dependent on a specific type of enhancement, thereby reducing the risk of overfitting. At the same time, the combination of multiple random data enhancements and their quantity and form allows for a more detailed simulation of possible shooting situations under different lighting conditions and altitudes, thereby improving the accuracy of insulation material warnings.

[0118] The secondary image and the corresponding material temperature, temperature difference, humidity, external parameter evaluation value and material electrical characteristic data are converted into feature vectors, and the material aging degree value is used as the label corresponding to the feature vector. The material aging degree value is set by the material expert as an integer of 1-5 according to the insulating material aging image corresponding to the secondary image. The higher the value, the higher the aging degree, and the more dangerous the insulating material is at this time. Each group of feature vectors and the label corresponding to each group of feature vectors are constructed as a sample, and multiple samples are collected to construct a data set; the data set is divided into a training set, a validation set and a test set, wherein the training set accounts for 60% of the data set, and the validation set and the test set each account for 20% of the data set;

[0119] The training set is used as the input of the machine learning model, and the machine learning model uses the real-time material aging degree value as the output; the historical material aging degree value corresponding to a set of real-time feature vectors is used as the prediction target, and minimizing the machine learning model loss function value is used as the training target; when the machine learning model loss function value is less than or equal to the preset target loss value, the training is stopped;

[0120] The machine learning model loss function is the mean square error; the mean square error is calculated by Minimize to train the model; in the loss function, MSE is the loss function value, i is the feature vector group number; u is the number of feature vector groups; y i is the label corresponding to the i-th group of feature vectors, The material aging degree value predicted by the i-th group of feature vectors;

[0121] The machine learning model is a deep neural network model.

[0122] Compare the results output by the machine learning model with the preset aging threshold, and perform early warning processing when it is greater than or equal to the preset aging threshold;

[0123] Early warning processing includes arranging staff to replace insulation materials, etc.

[0124] Example 1 monitors the aging images of insulating materials in real time, and uses a machine learning model to predict the degree of aging in combination with environmental parameters and material electrical property data. Example 1 collects aging images and performs preprocessing, including denoising, contrast enhancement and standardization. Perform data enhancement to simulate different lighting and environmental conditions. Use deep neural network model training to predict the degree of aging in real time. When the predicted result reaches the aging threshold, an early warning is triggered and corresponding processing is performed. The advantage of this solution is that it comprehensively considers multi-dimensional environmental factors and material properties, thereby improving the accuracy and reliability of the prediction. At the same time, it enhances the generalization ability of the model for different environmental conditions, reduces the risk of overfitting, and at the same time, the real-time monitoring and early warning mechanism ensures the safety of insulating materials and the stable operation of equipment.

[0125] Example 2

[0126] See also Figure 2 As shown, for the parts not described in detail in this embodiment, please refer to the description of Embodiment 1. A power equipment condition monitoring system based on image analysis is provided, and the system includes: each module is connected through a wired and / or wireless network.

[0127] A data collection module collects aging images of insulating materials, and at the same time collects environmental parameters and material electrical characteristic data corresponding to the aging images of insulating materials; environmental parameters include material temperature, temperature difference, humidity, altitude, ultraviolet intensity, and air quality index evaluation value; the temperature measured at the edge of a circle with a radius of 10 meters and a distance from the location of the insulating material as the center is marked as the environmental impact temperature, and the temperature difference is the difference between the environmental impact temperature and the material temperature; the air quality index evaluation value is obtained by the weighted sum of the air quality index; the material electrical characteristic data include surface resistance and partial discharge amplitude;

[0128] A calculation module calculates the external parameter evaluation value based on the altitude, ultraviolet intensity and air quality index evaluation value;

[0129] A preprocessing module is used to preprocess the insulation material aging image;

[0130] The data enhancement module marks the preprocessed insulation material aging image as a primary image; performs data enhancement based on the primary image; data enhancement includes: illumination transformation, contrast adjustment and noise injection; illumination transformation includes: color temperature adjustment, brightness change and directional light source change; contrast adjustment includes contrast optimization and local contrast enhancement; noise injection includes Gaussian noise, salt and pepper noise and Poisson noise;

[0131] The tuning module sets corresponding numbers for the eight forms of data enhancement, including color temperature adjustment, brightness change, directional light source change, contrast optimization, local contrast enhancement, Gaussian noise, salt and pepper noise, and Poisson noise; sets enhancement selection parameters and enhancement quantity parameters; obtains the enhancement quantity parameter before each primary image is input into the machine learning model, and the enhancement quantity parameter randomly generates one of the integers from 1 to 8; represents the number of data enhancement forms to be performed; at the same time, the enhancement selection parameter randomly generates one of the integers from 1 to 8, representing the number of data enhancement forms to be performed; processes the primary image according to the data enhancement form corresponding to the number; at the same time, determines whether the requirements of the enhancement quantity parameter are met, and if not, regenerates the enhancement selection parameter; selects the next data enhancement form to be executed; when the data enhancement form is repeated, regenerates it until the number of data enhancement forms meets the requirements of the enhancement quantity parameter, and marks the primary image at this time as a secondary image;

[0132] The model training module trains a machine learning model to predict the material aging degree based on the secondary image and the corresponding material temperature, temperature difference, humidity, external parameter evaluation value and material electrical characteristic data for early warning processing.

[0133] Example 3

[0134] See also Figure 3 As shown, according to another aspect of the present application, an electronic device 500 is also provided. The electronic device 500 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, the above-mentioned power equipment condition monitoring system and method based on image analysis can be executed.

[0135] The method or system according to the embodiment of the present application can also be used by Figure 3 The electronic device shown in the figure is implemented by the electronic device architecture. Figure 3 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a read-only memory (ROM) 503, a random access memory (RAM) 504, a communication port 505 connected to a network, an input / output component 506, a hard disk 507, etc. The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, may store a method for monitoring the condition of an electric power device based on image analysis provided by the present application. Furthermore, the electronic device 500 may also include a user interface 508. Of course, Figure 3 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 3 One or more components of an electronic device are shown.

[0136] Example 4

[0137] See also Figure 4 As shown, a computer-readable storage medium 600 according to an embodiment of the present application is shown. Computer-readable instructions are stored on the computer-readable storage medium 600. When the computer-readable instructions are executed by the processor, a power equipment condition monitoring system and method based on image analysis according to an embodiment of the present application described with reference to the above figures can be executed. The storage medium 600 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0138] In addition, according to the embodiments of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided by the present application. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0139] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through a wired network or a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0140] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0141] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0142] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a system and method for monitoring the condition of power equipment based on image analysis. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0143] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0144] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0145] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0146] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for monitoring the condition of electric power equipment based on image analysis, characterized in that: include: Collect insulation material aging images, and collect environmental parameters and material electrical characteristic data corresponding to the insulation material aging images; environmental parameters include material temperature, temperature difference, humidity, altitude, ultraviolet intensity, and air quality index evaluation value; the temperature measured at the edge of a circle with a radius of 10 meters and a distance from the location of the insulation material as the center is marked as the environmental impact temperature, and the temperature difference is the difference between the environmental impact temperature and the material temperature; the air quality index evaluation value is obtained by the weighted sum of the air quality index; the material electrical characteristic data include surface resistance and partial discharge amplitude; The external parameter evaluation value is calculated based on the altitude, ultraviolet intensity and air quality index evaluation value; the calculation process of the external parameter evaluation value is as follows: Where: WJ is the external parameter evaluation value, H is the altitude, ZW is the ultraviolet intensity, KQ is the air quality index; Preprocessing insulation material aging images; Marking the pre-processed insulation material aging image as a first-level image; Perform data augmentation based on primary images; Data enhancement includes: illumination transformation, contrast adjustment and noise injection; illumination transformation includes: color temperature adjustment, brightness change and directional light source change; contrast adjustment includes contrast optimization and local contrast enhancement; noise injection includes Gaussian noise, salt and pepper noise and Poisson noise; Set corresponding numbers for eight forms of data enhancement: color temperature adjustment, brightness change, directional light source change, contrast optimization, local contrast enhancement, Gaussian noise, salt and pepper noise, and Poisson noise; Set the enhancement selection parameter and the enhancement quantity parameter; before each first-level image is input into the machine learning model, obtain the enhancement quantity parameter, and randomly generate one of the integers from 1 to 8 for the enhancement quantity parameter; represent the number of data enhancement forms to be performed; at the same time, the enhancement selection parameter randomly generates one of the integers from 1 to 8, representing the number of data enhancement forms to be performed; process the first-level image according to the data enhancement form corresponding to the number; at the same time, determine whether the requirements of the enhancement quantity parameter are met, and if not, regenerate the enhancement selection parameter; select the next data enhancement form to be executed; when the data enhancement form is repeated, regenerate it until the number of data enhancement forms meets the requirements of the enhancement quantity parameter, and mark the first-level image at this time as a second-level image; Based on the secondary images and the corresponding material temperature, temperature difference, humidity, external parameter evaluation values ​​and material electrical characteristics data, a machine learning model is trained to predict the material aging degree value for early warning processing.

2. The method for monitoring the condition of electric power equipment based on image analysis according to claim 1, characterized in that: Collect UVA and UVB intensities; UV intensity is obtained by adding UVA intensity multiplied by 0.8 and UVB intensity multiplied by 0.

2.

3. The method for monitoring the condition of electric power equipment based on image analysis according to claim 2, characterized in that: The pre-processing comprises: Denoising: Use non-local means or BM3D algorithm to reduce the noise of insulation material aging images; Contrast enhancement: Adaptive histogram equalization is applied to improve the contrast of the insulation material aging image; Standardization: The pixel values ​​of the insulation material aging image are standardized to the range of [0,1].

4. The method for monitoring the condition of electric power equipment based on image analysis according to claim 3 is characterized in that: Color temperature adjustments include: Convert the primary image to CIE XYZ color space; Apply Bradford transformation matrix to simulate the adaptation of human eyes to different light sources; Calculate the transformation matrix according to the color temperature: According to the color temperature of the image, calculate a Bradford transformation matrix corresponding to a specific color temperature; the specific color temperature is set by material science experts and electrical experts, ranging from 4000K to 6000K; Convert the transformed XYZ value back to RGB color space; Convert the XYZ value adjusted by color temperature back to RGB color space through the inverse transformation matrix; The brightness change includes:

1. Color space conversion: convert the image to HSV or LAB color space; 2. Brightness channel adjustment: linear or nonlinear transformation of V channel or L channel; 3. Gamma correction: Use gamma correction to simulate the nonlinear perception of brightness changes by the human eye; 4. Color space conversion: convert the modified image back to the original color space; The process of directional light source change includes: creating a gradient map with the same size as the first-level image; using trigonometric functions to generate the lighting pattern; alpha blending or multiplication of the gradient map with the original image: combining the lighting information in the gradient map with the pixel values ​​of the original image to simulate the lighting effect.

5. The method for monitoring the condition of electric power equipment based on image analysis according to claim 4 is characterized in that: The contrast optimization process includes: a. Using histogram analysis to determine the contrast range of the image: Analyze the histogram of the image to understand the distribution of bright and dark pixels in order to adjust the contrast; b. Apply gamma correction; c. Use linear contrast stretching to enhance the contrast; d. Use the S-curve of the sigmoid function for nonlinear contrast adjustment.

6. The method for monitoring the condition of electric power equipment based on image analysis according to claim 5, characterized in that: The secondary image and the corresponding material temperature, temperature difference, humidity, external parameter evaluation value and material electrical characteristic data are converted into feature vectors, the material aging degree value is used as the label corresponding to the feature vector, and the material aging degree value is an integer of 1-5. Each set of feature vectors and the label corresponding to each set of feature vectors are constructed as a sample, and multiple samples are collected to construct a data set; the data set is divided into a training set, a validation set and a test set, wherein the training set accounts for 60% of the data set, and the validation set and the test set each account for 20% of the data set; Using the training set as input to a machine learning model, the machine learning model outputting a real-time material aging value; The historical material aging degree values ​​corresponding to a set of real-time feature vectors are used as prediction targets, and the minimization of the machine learning model loss function value is used as the training target; Stop training when the machine learning model loss function value is less than or equal to the preset target loss value; The machine learning model loss function is the mean square error; the mean square error is calculated by Minimize to train the model; in the loss function, MSE is the loss function value, i is the feature vector group number; u is the number of feature vector groups; y i is the label corresponding to the i-th group of feature vectors, The material aging degree value predicted by the i-th group of feature vectors; The machine learning model is a deep neural network model; the result output by the machine learning model is compared with a preset aging threshold, and an early warning process is performed when the result is greater than or equal to the preset aging threshold; Early warning measures include arranging staff to replace insulation materials.

7. An electric power equipment condition monitoring system based on image analysis, characterized in that: A data collection module collects aging images of insulating materials, and at the same time collects environmental parameters and material electrical characteristic data corresponding to the aging images of insulating materials; environmental parameters include material temperature, temperature difference, humidity, altitude, ultraviolet intensity, and air quality index evaluation value; the temperature measured at the edge of a circle with a radius of 10 meters and a distance from the location of the insulating material as the center is marked as the environmental impact temperature, and the temperature difference is the difference between the environmental impact temperature and the material temperature; the air quality index evaluation value is obtained by the weighted sum of the air quality index; the material electrical characteristic data include surface resistance and partial discharge amplitude; The calculation module calculates the external parameter evaluation value based on the altitude, ultraviolet intensity and air quality index evaluation value; the calculation process of the external parameter evaluation value is as follows: Where: WJ is the external parameter evaluation value, H is the altitude, ZW is the ultraviolet intensity, KQ is the air quality index; A preprocessing module is used to preprocess the insulation material aging image; A data enhancement module marks the preprocessed insulation material aging image as a primary image; Perform data augmentation based on primary images; Data enhancement includes: illumination transformation, contrast adjustment and noise injection; illumination transformation includes: color temperature adjustment, brightness change and directional light source change; contrast adjustment includes contrast optimization and local contrast enhancement; noise injection includes Gaussian noise, salt and pepper noise and Poisson noise; The tuning module sets corresponding numbers for the eight forms of data enhancement, including color temperature adjustment, brightness change, directional light source change, contrast optimization, local contrast enhancement, Gaussian noise, salt and pepper noise, and Poisson noise; sets enhancement selection parameters and enhancement quantity parameters; obtains the enhancement quantity parameter before each primary image is input into the machine learning model, and the enhancement quantity parameter randomly generates one of the integers from 1 to 8; represents the number of data enhancement forms to be performed; at the same time, the enhancement selection parameter randomly generates one of the integers from 1 to 8, representing the number of data enhancement forms to be performed; processes the primary image according to the data enhancement form corresponding to the number; at the same time, determines whether the requirements of the enhancement quantity parameter are met, and if not, regenerates the enhancement selection parameter; selects the next data enhancement form to be executed; when the data enhancement form is repeated, regenerates it until the number of data enhancement forms meets the requirements of the enhancement quantity parameter, and marks the primary image at this time as a secondary image; The model training module trains a machine learning model to predict the material aging degree based on the secondary image and the corresponding material temperature, temperature difference, humidity, external parameter evaluation value and material electrical characteristic data for early warning processing.

8. An electronic device, characterized in that: include: A processor and a memory, wherein the memory stores a computer program that can be called by the processor; The processor executes the method for monitoring the condition of electric power equipment based on image analysis as described in any one of claims 1 to 6 by calling the computer program stored in the memory.

9. A computer-readable storage medium, characterized in that: Instructions are stored, and when the instructions are executed on a computer, the computer is caused to execute a method for monitoring the condition of electric power equipment based on image analysis as claimed in any one of claims 1 to 6.

Images