Electric power fitting corrosion degree evaluation method and device, terminal equipment and storage medium

Abstract

The invention discloses an electric power fitting corrosion degree assessment method and device, terminal equipment and a storage medium. The method comprises the following steps: inputting a to-be-identified image of a to-be-assessed electric power fitting into a preset fitting corrosion degree assessment model to obtain a corrosion degree assessment result; the training of the fitting corrosion degree evaluation model comprises the following steps: performing corrosion degree marking according to a corrosion RGB interval, an RGB numerical value, texture roughness and a contrast ratio of an image sample of the electric power fitting to obtain a first target sample; training a to-be-trained hardware fitting corrosion degree evaluation model according to the first target sample and the current model parameters to obtain a predicted corrosion degree; obtaining a trained hardware corrosion degree evaluation model under the condition of loss value convergence; taking the updated current model parameter as the current model parameter of the next training under the condition that the loss value is not converged; the current model parameter trained for the first time is an initial model parameter. According to the invention, the safety and efficiency of corrosion degree evaluation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and particularly to a method, device, terminal device and storage medium for evaluating the corrosion degree of electric power fittings. Background Art

[0002] As an important component of the transmission line, the insulator string and its supporting fittings undertake key functions such as mechanical fixation, electrical insulation and load transfer. During actual operation, due to being long-term exposed to complex and changeable natural environments, being subjected to the continuous effects of ultraviolet radiation, air corrosion, rain erosion and temperature and humidity changes, and coupled with the superimposed effects of dynamic loads such as mechanical vibration and external forces, the surface of the fittings is corroded, thus forming a hidden danger of discharge, seriously affecting the normal operation of the transmission line.

[0003] In order to ensure the safety of the transmission line, it is necessary to regularly detect the corrosion condition of the electric power fittings in the transmission line to evaluate the corrosion degree of the electric power fittings. The existing evaluation methods mainly rely on manual work by climbing the tower during planned power outages or performing live high-altitude work with the aid of insulating tools. The frequency of climbing the tower for manual inspection to evaluate the corrosion of electric power fittings is high, posing a great personal risk to the operators. At the same time, some inspection routes are in special terrains, so the manual inspection is difficult, and the project establishment cycle of the maintenance items is also long, resulting in low efficiency in evaluating the corrosion degree of electric power fittings. Summary of the Invention

[0004] The embodiments of the present invention provide a method, device, terminal device and storage medium for evaluating the corrosion degree of electric power fittings, which can effectively solve the problems of high risk and low efficiency in the existing manual evaluation of the corrosion degree of electric power fittings.

[0005] An embodiment of the present invention provides a method for evaluating the corrosion degree of electric power fittings, including:

[0006] Obtaining an image to be recognized of the electric power fittings to be evaluated;

[0007] Inputting the image to be recognized into a preset evaluation model for the corrosion degree of the fittings to perform detection, and obtaining an evaluation result of the corrosion degree of the electric power fittings to be evaluated;

[0008] Among them, the training of the evaluation model for the corrosion degree of the fittings includes:

[0009] Obtaining image samples of a plurality of electric power fittings, the rust RGB intervals corresponding to each image sample, and initial model parameters;

[0010] Determining the RGB values, texture roughness and contrast corresponding to each image sample according to the image samples;

[0011] Mark the corrosion degree of each image sample according to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to each image sample, and obtain the first target sample after marking;

[0012] Train the fitting corrosion degree evaluation model to be trained according to the first target sample and the current model parameters to obtain the predicted corrosion degree;

[0013] Calculate the loss value of the preset loss function according to the predicted corrosion degree and the first target sample;

[0014] When the loss value converges, obtain the trained fitting corrosion degree evaluation model;

[0015] When the loss value does not converge, update the current model parameters according to the loss value, and use the updated current model parameters as the current model parameters for the next training; among them, the current model parameters at the first training are the initial model parameters.

[0016] Further, obtaining the rust RGB interval corresponding to each image sample includes:

[0017] Determine the fitting material corresponding to each image sample according to the image samples of several fitting;

[0018] Determine the color change after rusting of each fitting material according to the fitting material;

[0019] Determine the rust RGB interval corresponding to each image sample according to the color change after rusting of each fitting material.

[0020] Further, marking the corrosion degree of each image sample according to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to each image sample, and obtaining the first target sample after marking, includes:

[0021] Judge whether the RGB value is within the corresponding rust RGB interval according to the RGB value;

[0022] If so, determine that the image sample is the first image sample used to represent that the fitting is a corroded fitting;

[0023] If not, determine that the image sample is the second image sample used to represent that the fitting is a non-corroded fitting;

[0024] Perform image segmentation on the first image sample to obtain the proportion of the corrosion area of the first image sample;

[0025] Perform weighted calculation based on the corrosion area ratio, the RGB values, the texture roughness, the contrast, and a preset weight value to obtain the target corrosion value of the first image sample;

[0026] Traverse according to the target corrosion value in a preset corrosion degree mapping table to obtain the target corrosion degree of the first image sample;

[0027] Mark the first image sample according to the target corrosion degree, and mark the corrosion degree of the second image sample as 0 to obtain the marked first target sample;

[0028] Among them, the preset corrosion degree mapping table is used to store the corresponding relationship between the corrosion value and the corrosion degree of each electric power fitting.

[0029] Further, it also includes:

[0030] Obtain the corresponding relationship between the corrosion degree of each electric power fitting and the number of paint spraying covers;

[0031] Mark the number of paint spraying covers for the image sample according to the target corrosion degree and the corresponding relationship between the corrosion degree of each electric power fitting and the number of paint spraying covers to obtain the marked second target sample;

[0032] Train the trained electric power fitting corrosion degree evaluation model according to the second target sample to obtain the target model.

[0033] Further, it also includes:

[0034] Input the image to be recognized into the target model for detection to obtain the target number of paint spraying covers of the image to be recognized;

[0035] Determine the RGB values, texture roughness, and contrast corresponding to each image to be recognized according to the image to be recognized;

[0036] Determine the surface to be sprayed of the electric power fitting to be evaluated according to the RGB values, texture roughness, and contrast corresponding to each image to be recognized;

[0037] Generate a control instruction according to the target number of paint spraying covers and the surface to be sprayed, and send the control instruction to the unmanned aerial vehicle, so that the unmanned aerial vehicle performs a spraying operation on the corresponding electric power fitting to be evaluated according to the control instruction according to the target number of paint spraying covers on the surface to be sprayed.

[0038] Further, it also includes:

[0039] When a new material of electric power fittings is detected, according to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to the new material of electric power fittings, mark the corrosion degree of each corresponding new image to obtain the marked new samples;

[0040] Update the first target samples according to the marked new samples, and update the fitting corrosion degree evaluation model according to the updated first target samples.

[0041] Further, according to the image samples, determine the RGB value, texture roughness, and contrast corresponding to each image sample, including:

[0042] Traverse the pixels in the image samples according to the image samples to determine the corresponding RGB values;

[0043] Convert the image samples into grayscale images according to the preset gray-level co-occurrence matrix;

[0044] Calculate the corresponding texture roughness according to the grayscale image and the preset gray-level co-occurrence matrix;

[0045] Calculate the image gradient amplitude according to the image samples and the preset Laplacian operator;

[0046] Calculate the variance of the image gradient amplitude according to the image gradient amplitude, and use the variance of the image gradient amplitude as the contrast corresponding to the image samples.

[0047] As an improvement of the above solution, another embodiment of the present invention correspondingly provides an electric power fitting corrosion degree evaluation device, including:

[0048] An electric power fitting image acquisition module, configured to acquire an image to be recognized of an electric power fitting to be evaluated;

[0049] A corrosion degree evaluation module, configured to input the image to be recognized into a preset fitting corrosion degree evaluation model for detection to obtain an evaluation result of the corrosion degree of the electric power fitting to be evaluated;

[0050] Wherein, it further includes an evaluation model training module;

[0051] The evaluation model training module is configured to train the fitting corrosion degree evaluation model, including:

[0052] A sample data acquisition unit, configured to acquire image samples of a plurality of electric power fittings, the rust RGB interval corresponding to each image sample, and initial model parameters;

[0053] A sample feature determination unit, configured to determine the RGB value, texture roughness, and contrast corresponding to each image sample according to the image samples;

[0054] An image sample marking unit, configured to mark the corrosion degree of each image sample according to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to each image sample, so as to obtain a first target sample after marking;

[0055] A corrosion degree prediction unit, configured to train a fitting corrosion degree evaluation model for fittings to be trained according to the first target sample and current model parameters, so as to obtain a predicted corrosion degree;

[0056] A loss function calculation unit, configured to calculate a loss value of a preset loss function according to the predicted corrosion degree and the first target sample;

[0057] A model determination unit, configured to obtain a trained fitting corrosion degree evaluation model when the loss value converges;

[0058] A model update unit, configured to update the current model parameters according to the loss value when the loss value does not converge, and use the updated current model parameters as the current model parameters for the next training; wherein, the current model parameters at the first training are initial model parameters.

[0059] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the method for evaluating the corrosion degree of power fittings described in the above embodiment is implemented.

[0060] Another embodiment of the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program. When the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the method for evaluating the corrosion degree of power fittings described in the above embodiment.

[0061] By implementing the present invention, at least the following beneficial effects are achieved:

[0062] The present invention provides a method, device, terminal device and storage medium for evaluating the corrosion degree of electrical fittings. The method can obtain an image to be recognized of the electrical fitting to be evaluated; input the image to be recognized into a preset evaluation model for the corrosion degree of the fitting to perform detection, and obtain an evaluation result of the corrosion degree of the electrical fitting to be evaluated; wherein, the training of the evaluation model for the corrosion degree of the fitting includes: obtaining a plurality of image samples of electrical fittings, the rust RGB interval corresponding to each image sample, and initial model parameters; determining the RGB value, texture roughness and contrast corresponding to each image sample according to the image sample; performing corrosion degree marking on each image sample according to the rust RGB interval, RGB value, texture roughness and contrast corresponding to each image sample to obtain a first target sample after marking; training the evaluation model for the corrosion degree of the fitting to be trained according to the first target sample and the current model parameters to obtain a predicted corrosion degree; calculating the loss value of a preset loss function according to the predicted corrosion degree and the first target sample; when the loss value converges, obtaining a trained evaluation model for the corrosion degree of the fitting; when the loss value does not converge, updating the current model parameters according to the loss value, and using the updated current model parameters as the current model parameters for the next training; wherein, the current model parameters at the first training are the initial model parameters. By obtaining the image to be recognized of the electrical fitting to be evaluated and inputting it into a preset evaluation model for the corrosion degree of the fitting, automatic detection of the corrosion degree is realized, manual participation is reduced, the risk of manual tower climbing operation or live working at height is reduced, the difficulty of manual inspection is avoided through model detection, the inspection cost is reduced, and the evaluation efficiency of the corrosion degree of the electrical fitting is improved; the model is trained based on features such as the corrosion RGB interval, RGB value, texture roughness and contrast of the image sample, integrating multi-dimensional indicators, and can more accurately identify the corrosion degree of the electrical fitting to be recognized; the evaluation model for the corrosion degree of the fitting is trained by using each first target sample after marking, and the image to be recognized of the electrical fitting to be evaluated is automatically processed based on the evaluation model for the corrosion degree of the fitting without manual operation, further improving the accuracy of the evaluation. Description of the Drawings

[0063] Figure 1 is a schematic flowchart of a method for evaluating the corrosion degree of electrical fittings provided by an embodiment of the present invention;

[0064] Figure 2 is a schematic diagram of an electrical fitting to be evaluated provided by an embodiment of the present invention;

[0065] Figure 3 is a schematic flowchart of the training process of the evaluation model for the corrosion degree of the fitting provided by an embodiment of the present invention;

[0066] Figure 4It is a schematic diagram of the color change after corrosion of the power fittings to be evaluated provided by an embodiment of the present invention;

[0067] Figure 5 It is a schematic structural diagram of an evaluation device for the corrosion degree of power fittings provided by an embodiment of the present invention;

[0068] Figure 6 It is a schematic structural diagram of an evaluation model training module provided by an embodiment of the present invention. Detailed implementation manners

[0069] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0070] See Figure 1 , which is a schematic flowchart of a method for evaluating the corrosion degree of power fittings provided by an embodiment of the present invention, including:

[0071] S1. Obtain the image to be recognized of the power fittings to be evaluated;

[0072] Specifically, in response to the corrosion degree detection instruction for the power fittings to be evaluated, an image of the power fittings to be evaluated is collected as the image to be recognized. The corrosion degree detection instruction can be triggered by relevant staff according to actual needs. Receive the FPV screen (first-person perspective screen) collected by any unmanned aerial vehicle as the image to be recognized. Directly use the unmanned aerial vehicle to collect the image to be recognized of the power fittings to be evaluated and perform real-time detection based on the collected image to be recognized, effectively improving the detection efficiency.

[0073] In this embodiment, the power fittings to be recognized include, but are not limited to: triangular link plates, fittings, and suspension rings, as Figure 2 (a) and Figure 2 (b) shown, which are schematic diagrams of the power fittings to be evaluated.

[0074] S2. Input the image to be recognized into a preset evaluation model for the corrosion degree of fittings to perform detection, and obtain the evaluation result of the corrosion degree of the power fittings to be evaluated;

[0075] Among them, the training of the evaluation model for the corrosion degree of fittings, as Figure 3 shown, includes:

[0076] S301. Obtain image samples of a number of power fittings, the rust RGB intervals corresponding to each image sample, and initial model parameters;

[0077] Specifically, the rust RGB interval corresponding to each image sample is obtained, including:

[0078] According to a number of image samples of electric fittings, determine the material of the electric fitting corresponding to each image sample;

[0079] According to the material of the electrical fittings, determine the color change of each electrical fitting material after rusting;

[0080] According to the color change of each electrical fitting material after rusting, the rust RGB interval corresponding to each image sample is determined.

[0081] Preferably, the image samples of several electrical fittings are collected by a high-resolution camera. In this embodiment, the high-resolution camera has a high resolution and can record more details to make the picture clearer. In addition, high resolution means that there are enough remaining pixels even after later cropping, so it is convenient for subsequent model processing. The rust RGB interval corresponding to each image sample is the interval of the three primary colors of rust (Red, Green, and Blue). The rust RGB interval refers to the color change interval of the electrical fittings after rusting, such as Figure 4 (a) and Figure 4 (b) As shown. Taking the triangular connecting plate and hanging ring as an example, Figure 4 (a) is the color state when it is not corroded. Figure 4 (b) is the color state after rusting. It can be seen that the color of the power fittings will change significantly after rusting, and the corresponding RGB value will also change significantly. According to the color change of the power fittings to be evaluated of different materials after rusting, the corresponding rust RGB interval is configured in advance to assist in the subsequent automatic detection of whether the power fittings are rusted.

[0082] For example, for steel and cast iron, the metal surface is dark when it starts to rust. Light rust is dark gray, and further development will turn into brown or brownish yellow. Severe rust will appear as brown or brown scars or even rust pits. Moreover, after scraping off the rust products, the bottom is dark gray with irregular edges. The colors of rust products include yellow (Fe(OH) 3 ), black (Fe 3 O 4 ), brown (FeO(OH)), red (Fe 2 O 3 ), dark brown (FeCl 3 ), dark green (FeCl 2 ) etc. For copper and copper alloys, copper rusts green, orange or brown. The rust products of copper include black (CuO, CuS), orange (Cu 2 O), green (CuCl 2 、Cu(OH) 2 CuCO 3) etc. For aluminum and aluminum alloys, grayish white spots appear in the early stage, and grayish white rust products appear after development. After scraping off the rust products, pits appear at the bottom. The rust products of aluminum include white (Al 2 O 3 、Al(OH) 3 、AlCl 3 ).

[0083] In a preferred embodiment of the present invention, the material of the power fittings includes but is not limited to: stainless steel, cast iron, galvanized cast iron and aluminum alloy; when not corroded, the surface color of the fittings is mostly silver-gray, and after being corroded, the color changes of different materials will be different. Stainless steel, cast iron, etc. will produce brown-red or brown rust; aluminum alloy may produce large areas of black with dense gray or white spots. It can be seen that because power fittings of different materials change color differently after rusting, this embodiment specifically configures the corresponding rust RGB interval according to the power fittings of different materials to improve the accuracy of detection.

[0084] S302, determining the RGB value, texture roughness and contrast corresponding to each image sample according to the image samples;

[0085] Specifically, according to the image samples, determining the RGB value, texture roughness and contrast corresponding to each image sample includes:

[0086] Traversing pixels in the image sample according to the image sample, determining corresponding RGB values;

[0087] Performing grayscale conversion on the image sample according to a preset grayscale co-occurrence matrix to obtain a grayscale image;

[0088] Calculating according to the grayscale image and a preset grayscale co-occurrence matrix to obtain the corresponding texture roughness;

[0089] Calculating an image gradient amplitude according to the image sample and a preset Laplace operator;

[0090] The image gradient amplitude variance is calculated according to the image gradient amplitude, and the image gradient amplitude variance is used as the contrast corresponding to the image sample.

[0091] In a preferred embodiment of the present invention, the OpenCV library can be used to read each sample image and the pixels in the image sample to obtain the RGB values of the electrical hardware at specific positions. The preset gray-level co-occurrence matrix can be used to perform gray-scale conversion on the image sample to obtain a gray-scale image, and then the preset gray-level co-occurrence matrix is used to calculate the texture roughness. The preset Laplacian operator in OpenCV can be used to calculate the image gradient magnitude, and then the variance of the image gradient magnitude is calculated according to the image gradient magnitude, and the variance of the image gradient magnitude is used as the contrast corresponding to the image sample. Or other algorithms can also be used to detect the RGB values, texture roughness, and contrast of the electrical hardware in each image sample, and the specific algorithms used in this embodiment are not limited.

[0092] S303. According to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to each image sample, mark the corrosion degree of each image sample to obtain the first target sample after marking;

[0093] Specifically, according to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to each image sample, marking the corrosion degree of each image sample to obtain the first target sample after marking includes:

[0094] According to the RGB value, determine whether the RGB value is within the corresponding rust RGB interval;

[0095] If so, determine that the image sample is the first image sample used to represent that the electrical hardware is a corroded hardware;

[0096] If not, determine that the image sample is the second image sample used to represent that the electrical hardware is a non-corroded hardware;

[0097] Perform image segmentation on the first image sample to obtain the proportion of the corrosion area of the first image sample;

[0098] Perform weighted calculation according to the proportion of the corrosion area, the RGB value, the texture roughness, the contrast, and the preset weight value to obtain the target corrosion value of the first image sample;

[0099] Traverse according to the target corrosion value in the preset corrosion degree mapping table to obtain the target corrosion degree of the first image sample;

[0100] Mark the first image sample according to the target corrosion degree, and mark the corrosion degree of the second image sample as 0 to obtain the first target sample after marking;

[0101] Among them, the preset corrosion degree mapping table is used to store the corresponding relationship between the corrosion value and the corrosion degree of each electrical hardware.

[0102] In a preferred embodiment of the present invention, according to the RGB values, it is determined whether the RGB values are within the corresponding rust RGB range. When the RGB values are within the corresponding rust RGB range, it indicates that the corresponding electrical fitting is a corroded fitting, and then the image sample is determined as the first image sample for characterizing that the electrical fitting is a corroded fitting. When the RGB values are not within the corresponding rust RGB range, it indicates that the corresponding electrical fitting is a non-corroded fitting, and then the image sample is determined as the second image sample for characterizing that the electrical fitting is a non-corroded fitting. Then, the first image sample is segmented to obtain the proportion of the corroded area of the corroded fitting in the first image sample. A weighted calculation is performed based on the proportion of the corroded area, the RGB values, the texture roughness, the contrast, and a preset weight value to obtain the target corrosion value of the first image sample. Specifically, the preset weight value includes: the first weight corresponding to the proportion of the corroded area, the second weight corresponding to the RGB values, the third weight corresponding to the texture roughness, and the fourth weight corresponding to the contrast. A weighted calculation is performed based on the first weight, the second weight, the third weight, the fourth weight, the proportion of the corroded area, the RGB values, the texture roughness, and the contrast to obtain the target corrosion value of the first image sample. Then, according to the target corrosion value, a traversal is performed in a preset corrosion degree mapping table to obtain the target corrosion degree of the first image sample; the first image sample is marked according to the target corrosion degree, and the corrosion degree of the second image sample is marked as 0 to obtain the marked first target sample; wherein, the preset corrosion degree mapping table is used to store the corresponding relationship between the corrosion value and the corrosion degree of each electrical fitting. Taking stainless steel as an example, the target rust degree is divided into 5 levels, which is mainly reflected in the change of color depth. For the RGB values, if the R value changes the most with the corrosion degree, then only the R value can be used for calculation.

[0103] Through the above embodiments, it is possible to comprehensively detect the corrosion degree of electrical fittings by combining multi-dimensional indexes such as RGB values, texture roughness, contrast, and proportion of corroded area, mark the first target sample, improve the accuracy of model training, and improve the accuracy of detection.

[0104] S304. Train the electrical fitting corrosion degree evaluation model to be trained according to the first target sample and the current model parameters to obtain a predicted corrosion degree;

[0105] In a preferred embodiment of the present invention, the fitting corrosion degree evaluation model to be trained is an AI (Artificial Intelligence) vision model with a classification function. The AI vision model may include, but is not limited to: Convolutional Neural Networks (CNN), Residual Network (ResNet), YOLO (You Only Look Once) series models, etc.

[0106] Preferably, the first target samples are divided into a training set and a validation set according to a preset ratio. The training set is used to train the AI vision model, and the validation set is used to validate the trained model to ensure the training effect of the model. Moreover, during the training process, the model is continuously iterated with the marked training target so that the model can accurately process the samples.

[0107] S305. Calculate the loss value of a preset loss function according to the predicted corrosion degree and the first target samples;

[0108] Specifically, since the first target samples are marked with the corrosion degree, the loss value of the preset loss function can be calculated with the predicted corrosion degree.

[0109] S306. When the loss value converges, obtain the trained fitting corrosion degree evaluation model;

[0110] S307. When the loss value does not converge, update the current model parameters according to the loss value, and use the updated current model parameters as the current model parameters for the next training; wherein, the current model parameters at the first training are the initial model parameters.

[0111] Preferably, it further includes:

[0112] Obtain the corresponding relationship between the corrosion degree of each electrical fitting and the number of paint spraying times;

[0113] Mark the number of paint spraying times for the image samples according to the target corrosion degree and the corresponding relationship between the corrosion degree of each electrical fitting and the number of paint spraying times, and obtain the marked second target samples;

[0114] Train the trained fitting corrosion degree evaluation model according to the second target samples to obtain the target model.

[0115] In a preferred embodiment of the present invention, power fittings with different degrees of corrosion require different numbers of paint spraying coverages. As the degree level deepens, the number of spraying coverages of the antirust paint will vary. After determining the target corrosion degree according to the fitting corrosion degree evaluation model, mark the number of paint spraying coverages for the image sample to obtain the marked second target sample, and then train the trained fitting corrosion degree evaluation model according to the second target sample to obtain the target model. At this time, the target model can not only determine the target corrosion degree but also determine the target number of paint spraying coverages. Without manual judgment, it not only reduces the misjudgment rate but also improves the processing efficiency. Moreover, by automatically identifying the target rust degree of the item to be sprayed (the power fitting to be evaluated) and calculating the spraying dosage accordingly, it can effectively avoid paint waste.

[0116] Schematically, it further includes:

[0117] Input the image to be recognized into the target model for detection to obtain the target number of paint spraying coverages of the image to be recognized;

[0118] According to the image to be recognized, determine the RGB value, texture roughness, and contrast corresponding to each image to be recognized;

[0119] According to the RGB value, texture roughness, and contrast corresponding to each image to be recognized, determine the surface to be sprayed of the power fitting to be evaluated;

[0120] Generate a control instruction according to the target number of paint spraying coverages and the surface to be sprayed, and send the control instruction to the unmanned aerial vehicle, so that the unmanned aerial vehicle performs a spraying operation on the corresponding power fitting to be evaluated according to the target number of paint spraying coverages on the surface to be sprayed according to the control instruction.

[0121] In a preferred embodiment of the present invention, input the image to be recognized into the target model for detection to obtain the target number of paint spraying coverages of the corresponding power fitting to be evaluated in the image to be recognized. The target number of paint spraying coverages indicates how many times of paint spraying coverage are required for the power fitting to be evaluated after rusting (the target number of paint spraying coverages of a non-rusted power fitting is 0). According to the RGB value, texture roughness, and contrast corresponding to each image to be recognized, determine the surface to be sprayed of the power fitting to be evaluated. For example, the area where the RGB value, texture roughness, and contrast exceed the preset judgment threshold is determined as the surface to be sprayed of the power fitting to be evaluated. Then generate a control instruction according to the target number of paint spraying coverages and the surface to be sprayed, and send the control instruction to the unmanned aerial vehicle (the unmanned aerial vehicle carries spraying equipment), so that when the unmanned aerial vehicle reaches the preset distance from the corresponding power fitting to be evaluated according to the control instruction, it performs a spraying operation on the surface to be sprayed according to the target number of paint spraying coverages.

[0122] For example, after determining the target paint spraying coverage times of the image to be recognized using the model, the drone can be controlled to carry the spraying device and continuously approach the surface to be sprayed until it approaches the specified distance range (the distance between the nozzle and the actual surface to be sprayed should be within its effective spraying distance L), and when the item to be sprayed (the power fitting to be evaluated) is just within the FPV (First Person View) screen of the drone, the nozzle direction can be adjusted according to the position of the item to be sprayed in the screen through an electric joint (usually composed of an electric motor), and automatic spraying coverage can be carried out.

[0123] Schematically, it further includes:

[0124] When a new power fitting material is detected, according to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to the new power fitting material, each corresponding new image is marked with the degree of corrosion to obtain the marked new samples.

[0125] Update the first target samples according to the marked new samples, and update the power fitting corrosion degree evaluation model according to the updated first target samples.

[0126] In a preferred embodiment of the present invention, when it is detected that a new power fitting with a new material is put into use, configure the rust RGB interval, RGB value, texture roughness, and contrast corresponding to the new power fitting material; within the preset configuration duration, each corresponding new image is marked with the degree of corrosion to obtain the marked new samples; and the power fitting corrosion degree evaluation model is updated using the marked new samples. Among them, the preset configuration duration can be configured according to actual needs, such as half a year. Through the above embodiments, when a power fitting with a new material is put into use, the model can be updated by supplementing the first target samples, thereby ensuring the applicability of the model.

[0127] By implementing this embodiment, targeted treatment can be carried out according to the corrosion conditions of the operating fittings at special positions of different types such as strain towers and straight towers, preventing operation hazards caused by the corrosion of fittings, improving the operation and maintenance workload of personnel, simplifying the line maintenance operation process, and improving the safety of construction operations. Moreover, it reduces the manpower, material resources, and construction safety hazards consumed by operation and maintenance personnel for the corrosion of power fittings such as wire clamps and bolts, the cleaning and repair of insulator string pollution, effectively prevents and avoids the corrosion of power fittings during long-term operation, more comprehensively uses drones to replace manual operation procedures, avoids operators climbing to handle corroded fittings, reduces the personal risk of operations, and helps improve the safe and stable operation of the power grid, greatly reducing the operation and maintenance costs.

[0128] By obtaining the image to be recognized of the electric power fitting to be evaluated and inputting it into the preset evaluation model for the corrosion degree of the fitting, the automatic detection of the corrosion degree is realized, reducing manual participation and the risk of manual tower climbing operation or live working at height. Through model detection, the difficulties of manual inspection are avoided, the inspection cost is reduced, and the evaluation efficiency of the corrosion degree of the electric power fitting is improved; the model is trained based on features such as the corrosion RGB interval, RGB value, texture roughness, and contrast of the image samples, integrating multi-dimensional indicators, and can more accurately identify the corrosion degree of the electric power fitting to be recognized; the evaluation model for the corrosion degree of the fitting is trained using each first target sample after marking, and the image to be recognized of the electric power fitting to be evaluated is automatically processed based on the evaluation model for the corrosion degree of the fitting without manual operation, further improving the accuracy of the evaluation.

[0129] See Figure 5 , which is a schematic structural diagram of an evaluation device for the corrosion degree of an electric power fitting provided by an embodiment of the present invention, includes:

[0130] An electric power fitting image acquisition module for acquiring the image to be recognized of the electric power fitting to be evaluated;

[0131] A corrosion degree evaluation module for inputting the image to be recognized into a preset evaluation model for the corrosion degree of the fitting to perform detection and obtaining the evaluation result of the corrosion degree of the electric power fitting to be evaluated;

[0132] Preferably, it further includes an evaluation model training module for training the evaluation model for the corrosion degree of the fitting, Figure 6 which is a schematic structural diagram of the evaluation model training module provided by an embodiment of the present invention, and includes:

[0133] A sample data acquisition unit for acquiring image samples of several electric power fittings, the rust RGB interval corresponding to each image sample, and initial model parameters;

[0134] A sample feature determination unit for determining the RGB value, texture roughness, and contrast corresponding to each image sample according to the image sample;

[0135] An image sample marking unit for marking the corrosion degree of each image sample according to the rust RGB interval, RGB value, texture roughness, and contrast corresponding to each image sample to obtain the marked first target sample;

[0136] A corrosion degree prediction unit for training the evaluation model for the corrosion degree of the fitting to be trained according to the first target sample and the current model parameters to obtain the predicted corrosion degree;

[0137] A loss function calculation unit, configured to calculate a loss value of a preset loss function according to the predicted corrosion degree and the first target sample;

[0138] A model determination unit, configured to obtain a trained fitting corrosion degree evaluation model when the loss value converges;

[0139] A model update unit, configured to update current model parameters according to the loss value when the loss value does not converge, and use the updated current model parameters as the current model parameters for the next training; wherein, the current model parameters at the first training are initial model parameters.

[0140] The present invention provides a device for evaluating the corrosion degree of electrical fittings. According to the electrical fitting image acquisition module, an image to be recognized of the electrical fitting to be evaluated is acquired; through the corrosion degree evaluation module, the image to be recognized is input into a preset evaluation model for the corrosion degree of electrical fittings for detection, and an evaluation result of the corrosion degree of the electrical fitting to be evaluated is obtained; according to the sample data acquisition unit, image samples of a plurality of electrical fittings, the rust RGB interval corresponding to each image sample, and initial model parameters are acquired; in the sample feature determination unit, according to the image samples, the RGB values, texture roughness, and contrast corresponding to each image sample are determined; in the image sample marking unit, according to the rust RGB interval, RGB values, texture roughness, and contrast corresponding to each image sample, each image sample is marked with a corrosion degree to obtain a first target sample after marking; in the corrosion degree prediction unit, according to the first target sample and the current model parameters, the evaluation model for the corrosion degree of electrical fittings to be trained is trained to obtain a predicted corrosion degree; in the loss function calculation unit, according to the predicted corrosion degree and the first target sample, the loss value of a preset loss function is calculated; through the model determination unit, when the loss value converges, an evaluation model for the corrosion degree of electrical fittings that has been trained is obtained; through the model update unit, when the loss value does not converge, the current model parameters are updated according to the loss value, and the updated current model parameters are used as the current model parameters for the next training; wherein, the current model parameters at the first training are the initial model parameters. By acquiring the image to be recognized of the electrical fitting to be evaluated and inputting it into a preset evaluation model for the corrosion degree of electrical fittings, automatic detection of the corrosion degree is realized, manual participation is reduced, the risk of manual tower climbing operations or live working at high altitudes is reduced, the difficulties of manual inspection are avoided through model detection, the inspection cost is reduced, and the evaluation efficiency of the corrosion degree of electrical fittings is improved; the model is trained based on features such as the corrosion RGB interval, RGB values, texture roughness, and contrast of the image samples, comprehensively considering multi-dimensional indicators, and can more accurately identify the corrosion degree of the electrical fitting to be recognized; the evaluation model for the corrosion degree of electrical fittings is trained using each marked first target sample, and the image to be recognized of the electrical fitting to be evaluated is automatically processed based on the evaluation model for the corrosion degree of electrical fittings, without manual operation, further improving the accuracy of the evaluation.

[0141] It should be noted that the device embodiments described above are merely illustrative. 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 to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement this without creative work.

[0142] Those skilled in the art can clearly understand that for the sake of convenience and simplicity, the specific working process of the device described above can refer to the corresponding process in the foregoing method embodiment and will not be elaborated here.

[0143] Another embodiment of the present invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a method for evaluating the corrosion degree of electrical hardware as described in the above embodiment. The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0144] The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device and connects various parts of the entire terminal device through various interfaces and lines.

[0145] The memory can be used to store the computer program. By running or executing the computer program stored in the memory and invoking the data stored in the memory, the processor realizes various functions of the terminal device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include high-speed random access memory and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash memory device or other volatile solid-state storage devices.

[0146] Another embodiment of the present invention provides a computer-readable storage medium. The computer-readable storage medium includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the method for evaluating the corrosion degree of electrical hardware as described in the above embodiment.

[0147] The storage medium is a computer-readable storage medium. The computer program is stored in the computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the above various method embodiments. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0148] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A method for evaluating the corrosion degree of electric power fittings, characterized in that: include: Acquire the image to be identified of the electrical fitting to be evaluated; Inputting the image to be identified into a preset hardware corrosion degree assessment model for detection, and obtaining a corrosion degree assessment result of the power hardware to be assessed; The training of the hardware corrosion assessment model includes: Obtain image samples of several electrical fittings, the rust RGB interval corresponding to each image sample, and initial model parameters; According to the image samples, determining the RGB value, texture roughness and contrast corresponding to each image sample; Mark the corrosion degree of each image sample according to the rust RGB interval, RGB value, texture roughness and contrast corresponding to each image sample to obtain a marked first target sample; The hardware corrosion degree assessment model to be trained is trained according to the first target sample and the current model parameters to obtain a predicted corrosion degree; Calculating a loss value of a preset loss function according to the predicted corrosion degree and the first target sample; When the loss value converges, a trained hardware corrosion degree assessment model is obtained; When the loss value has not converged, the current model parameters are updated according to the loss value, and the updated current model parameters are used as the current model parameters for the next training; wherein, the current model parameters during the first training are the initial model parameters.

2. A method for evaluating the corrosion degree of electric power fittings according to claim 1, characterized in that: Get the rust RGB interval corresponding to each image sample, including: According to a number of image samples of electric fittings, determine the material of the electric fitting corresponding to each image sample; According to the material of the electrical fittings, determine the color change of each electrical fitting material after rusting; According to the color change of each electrical fitting material after rusting, the rust RGB interval corresponding to each image sample is determined.

3. A method for evaluating the corrosion degree of electric power fittings as claimed in claim 2, characterized in that: According to the rust RGB interval, RGB value, texture roughness and contrast corresponding to each image sample, the corrosion degree of each image sample is marked to obtain the marked first target sample, including: According to the RGB value, determining whether the RGB value is within a corresponding rust RGB interval; If yes, determining that the image sample is a first image sample used to characterize that the electrical fitting is a corroded fitting; If not, determining that the image sample is a second image sample used to characterize that the power fitting is a non-corrosive fitting; Performing image segmentation on the first image sample to obtain a corrosion area ratio of the first image sample; Performing weighted calculation according to the corrosion area ratio, the RGB value, the texture roughness, the contrast and a preset weight value to obtain a target corrosion value of the first image sample; Traversing a preset corrosion degree mapping table according to the target corrosion value to obtain a target corrosion degree of the first image sample; Marking the first image sample according to the target corrosion degree, and marking the corrosion degree of the second image sample as 0, to obtain a marked first target sample; The preset corrosion degree mapping table is used to store the corresponding relationship between the corrosion value and the corrosion degree of each electrical fitting.

4. A method for evaluating the corrosion degree of electric power fittings as claimed in claim 3, characterized in that: Also includes: Obtain the corresponding relationship between the corrosion degree of each electrical fitting and the number of times it has been painted; According to the target corrosion degree and the corresponding relationship between the corrosion degree of each electrical fitting and the number of times of paint coverage, the image sample is marked with the number of times of paint coverage to obtain a marked second target sample; The trained hardware corrosion degree assessment model is trained according to the second target sample to obtain a target model.

5. A method for evaluating the corrosion degree of electric power fittings as claimed in claim 4, characterized in that: Also includes: Inputting the image to be identified into the target model for detection, and obtaining the target paint coverage times of the image to be identified; According to the images to be identified, determining the RGB value, texture roughness and contrast corresponding to each image to be identified; Determine the spraying surface of the electrical fittings to be evaluated according to the RGB value, texture roughness and contrast corresponding to each image to be identified; A control instruction is generated according to the target paint coverage times and the surface to be sprayed, and the control instruction is sent to the drone, so that the drone performs a spraying operation on the corresponding electrical fitting to be evaluated according to the control instruction and the surface to be sprayed according to the target paint coverage times.

6. A method for evaluating the corrosion degree of electric power fittings as claimed in claim 5, characterized in that: Also includes: When a new material of electric fittings is detected, the corrosion degree of each corresponding new image is marked according to the rust RGB interval, RGB value, texture roughness and contrast corresponding to the new material of electric fittings, and the marked new samples are obtained; The first target sample is updated according to the marked newly added sample, and the hardware corrosion degree assessment model is updated according to the updated first target sample.

7. A method for evaluating the corrosion degree of electric power fittings as claimed in claim 1, characterized in that: According to the image samples, determining the RGB value, texture roughness and contrast corresponding to each image sample includes: Traversing pixels in the image sample according to the image sample, determining corresponding RGB values; Performing grayscale conversion on the image sample according to a preset grayscale co-occurrence matrix to obtain a grayscale image; Calculating according to the grayscale image and a preset grayscale co-occurrence matrix to obtain the corresponding texture roughness; Calculating an image gradient amplitude according to the image sample and a preset Laplace operator; The image gradient amplitude variance is calculated according to the image gradient amplitude, and the image gradient amplitude variance is used as the contrast corresponding to the image sample.

8. A device for evaluating the corrosion degree of electric power fittings, characterized in that: include: An electric fitting image acquisition module is used to acquire an image to be identified of the electric fitting to be evaluated; A corrosion degree assessment module is used to input the image to be identified into a preset hardware corrosion degree assessment model for detection, and obtain a corrosion degree assessment result of the power hardware to be assessed; Among them, it also includes an evaluation model training module; The assessment model training module is used for training the hardware corrosion assessment model, including: A sample data acquisition unit, used to acquire image samples of a number of electrical fittings, a corrosion RGB interval corresponding to each image sample, and initial model parameters; A sample feature determination unit, used to determine the RGB value, texture roughness and contrast corresponding to each image sample according to the image sample; An image sample marking unit is used to mark the corrosion degree of each image sample according to the rust RGB interval, RGB value, texture roughness and contrast corresponding to each image sample, and obtain a marked first target sample; A corrosion degree prediction unit, used for training a hardware corrosion degree assessment model to be trained according to the first target sample and current model parameters to obtain a predicted corrosion degree; A loss function calculation unit, used to calculate a loss value of a preset loss function according to the predicted corrosion degree and the first target sample; A model determination unit, used for obtaining a trained hardware corrosion degree assessment model when the loss value converges; A model updating unit is used to update the current model parameters according to the loss value when the loss value has not converged, and use the updated current model parameters as the current model parameters for the next training; wherein the current model parameters during the first training are the initial model parameters.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a method for evaluating the corrosion degree of electrical fittings as claimed in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute a method for evaluating the corrosion degree of electrical fittings as described in any one of claims 1 to 7.

Images