Sulfur hexafluoride barometer image identification reading method
By combining image annotation and target detection network YOLO with inspection robots, the automatic, efficient and accurate identification of sulfur hexafluoride pressure gauges was achieved, solving the problems of low efficiency and error-proneness of traditional manual reading and improving the identification effect.
Patent Information
- Application Number
- CN202511899867.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-16
- Publication Date
- 2026-04-14
Smart Images
Figure CN121861632A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of intelligent instrument reading technology, specifically relating to a method for reading images of a sulfur hexafluoride pressure gauge. Background Technology
[0002] Sulfur hexafluoride (SF6) gas is widely used in the insulation and arc-extinguishing systems of high-voltage electrical equipment (such as GIS switchgear and transformers). SF6 pressure gauges are used for long-term stability monitoring, real-time monitoring of SF6 gas leaks or pressure anomalies, and prevention of accidents caused by short circuits and arc faults. They play a crucial role in industrial safety monitoring and the new energy field. Currently, the main method for inspecting SF6 pressure gauges is manual inspection. However, traditional manual instrument reading methods are time-consuming, inefficient, and prone to errors. With the development of deep learning image processing technology, many vision-based pointer instrument reading methods have emerged. However, due to factors such as incomplete pointers and complex dials in SF6 pressure gauges, these image algorithms suffer from unsatisfactory recognition results and are prone to errors during intelligent reading, placing higher demands on deep learning image processing technology.
[0003] Therefore, researching an automated, efficient, and accurate intelligent image recognition and reading method for sulfur hexafluoride pressure gauges is of significant practical importance. Summary of the Invention
[0004] The purpose of this invention is to provide an image recognition reading method for sulfur hexafluoride pressure gauges, which can intelligently identify readings and realize automated, efficient and accurate reading recognition of sulfur hexafluoride pressure gauges.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for reading images of a sulfur hexafluoride pressure gauge, comprising the following steps: S1: Organize the pre-collected sulfur hexafluoride pressure gauge images and use image annotation software to annotate the dial, anchor points, and pointers in the images; S2: Input the labeled sulfur hexafluoride pressure gauge image data into the target detection network YOLO to train the YOLO model; S3: The inspection robot arrives at the inspection point according to the inspection route. The robot adjusts the camera to the pre-set position and pitch angle of the inspection point to take pictures and collect images of the sulfur hexafluoride pressure gauge at the inspection point, and collects a large field-of-view image containing the sulfur hexafluoride pressure gauge to be identified. S4: The recognition software module uses a trained YOLO model to recognize the large field-of-view image of the sulfur hexafluoride (SF6) pressure gauge, obtains the position information of the SF6 pressure gauge dial, and adjusts the azimuth and pitch angle of the camera according to the position information to center the SF6 pressure gauge in the image and ensure that the outer contour occupies an area of 1 / 4 to 1 / 2 of the entire image; based on the small field-of-view image of the SF6 pressure gauge to be recognized acquired by the camera; S5: The recognition software module uses the trained YOLO model to recognize the small field-of-view image and obtains the position information and confidence information of the dial, anchor point and pointer of the sulfur hexafluoride pressure gauge; S6: The recognition software module combines the prior information of the instruments at the inspection points to obtain accurate position information of the dial, anchor point and pointer; S7: The identification software module connects the anchor point and pointer center point of the sulfur hexafluoride pressure gauge obtained from S5 to form the actual position information of the pointer, and calculates the angle between the pointer and the zero mark by combining the prior information of the instrument. S8: Calculate the reading of the sulfur hexafluoride pressure gauge based on the prior information of the instruments at the inspection points and the included angle; Wherein: the prior information of the instrument is based on manual calibration.
[0006] The sulfur hexafluoride pressure gauge image recognition and reading method provided by the present invention also has the following technical features: in step S1, before image processing, the method further includes: cleaning the pre-collected sulfur hexafluoride pressure gauge images to remove duplicate images, missing images and useless images, and using an image annotation tool to annotate the cleaned images to annotate the dial, anchor point and pointer in the sulfur hexafluoride pressure gauge images. Before sending the labeled sulfur hexafluoride pressure gauge image data into the target detection network YOLO, the process also includes: dividing the shuffled labeled image dataset into training set, validation set and test set according to different proportions.
[0007] The sulfur hexafluoride pressure gauge image recognition reading method provided by this invention also has the following technical features: In step S2, a target detection network YOLO is first constructed. The network model specifically includes an input end, a backbone, a neck, and a prediction output end. The YOLO input end performs image preprocessing before inputting the model. Image preprocessing includes image resizing, RGB value normalization of image pixels, and image data enhancement. Image data enhancement increases the diversity of samples through image processing methods such as random cropping, scaling, rotation, brightness variation, adding Gaussian noise, and randomly stitching multiple images according to a certain ratio. The prediction output end is the detection head, which consists of convolutional layers, candidate box generation layers, prediction layers, and activation functions, and is used to generate the final target detection result. Its total loss function is calculated as follows:
[0008] in, For the total loss function, For classifying losses, For confidence loss, To locate the loss, For classification loss weights, For confidence loss weights, To determine the loss weights.
[0009] The sulfur hexafluoride pressure gauge image recognition reading method provided by the present invention also has the following technical features: in step 2, the hyperparameters of the network model are set, and the model is trained until convergence using the training set and validation set in the labeled dataset. Finally, the YOLO model weights with the best recognition effect are selected by testing on the test set.
[0010] The sulfur hexafluoride pressure gauge image recognition and reading method provided by the present invention also has the following technical features: in step S5, the prior information of the instrument at the inspection point includes the confidence level of the pointer, the confidence level of the pointer is the credibility of the recognition result, the confidence level threshold is set to 0.3, the detection results with a confidence level less than 0.3 are discarded, and the detection results with a confidence level greater than or equal to 0.3 are retained.
[0011] The sulfur hexafluoride pressure gauge image recognition and reading method provided by the present invention also has the following technical features: in step S6, the prior information of the instrument at the inspection point includes single dial information; the recognition results in step S5 are screened and grouped; invalid dials are arbitrated and deleted to obtain the final accurate dial position information.
[0012] The sulfur hexafluoride pressure gauge image recognition and reading method provided by the present invention also has the following technical features: in step S6, the prior information of the instrument at the inspection point also includes anchor points and pointers; the recognition results in step S5 are screened and grouped; invalid anchor points and pointers are arbitrated and deleted to obtain accurate anchor points and pointers.
[0013] The sulfur hexafluoride pressure gauge image recognition and reading method provided by the present invention also has the following technical features: S6 further includes an identification software module that combines the prior information of the instrument at the inspection point and groups them according to whether they are within the dial, so as to obtain accurate dial, anchor point and pointer.
[0014] Beneficial effects: (1) This invention provides an image recognition reading method for sulfur hexafluoride pressure gauges in power systems, which solves the problems of incomplete pointers and complex dials in sulfur hexafluoride pressure gauges, realizes automatic recognition and reading, and improves the efficiency and accuracy of reading; (2) This invention uses advanced deep learning methods to improve the universality and robustness of the algorithm; (3) Compared with the traditional manual reading method, the image recognition reading method for sulfur hexafluoride pressure gauges in power systems not only saves time and is more efficient, but also is less prone to omissions and errors. It has the advantages of simple operation, high degree of automation, high detection efficiency and accurate results, and is suitable for large-scale promotion and has a wide range of application prospects. Attached Figure Description
[0015] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 A flowchart of the method provided in the embodiments of the present invention; Figure 2 This is a large field-of-view image of a sulfur hexafluoride pressure gauge provided in an embodiment of the present invention; Figure 3 This is a large field-of-view image of the sulfur hexafluoride pressure gauge provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the recognition reading results of the large field-of-view image of the sulfur hexafluoride pressure gauge in an embodiment of the present invention. Detailed Implementation
[0017] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.
[0018] The following specific examples illustrate the implementation of this disclosure. Those skilled in the art can easily understand other advantages and effects of this disclosure from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. This disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this disclosure. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.
[0019] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using other structures and / or functionalities besides one or more of the aspects set forth herein.
[0020] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this disclosure. The drawings only show the components related to this disclosure and are not drawn according to the number, shape and size of the components in actual implementation. In actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.
[0021] Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details.
[0022] like Figure 1-4 As shown, this embodiment of the invention provides a method for reading images of a sulfur hexafluoride pressure gauge, including the following steps: S1: Organize the pre-collected sulfur hexafluoride pressure gauge images and use the image annotation tool [LableImg] to annotate the dial, anchor points, and pointers in the images; S2: Input the labeled sulfur hexafluoride pressure gauge image data into the target detection network YOLO to train the YOLO model; S3: The inspection robot arrives at the inspection point according to the inspection route. The robot adjusts the camera to the pre-set position and pitch angle of the inspection point to take pictures and collect images of the sulfur hexafluoride pressure gauge at the inspection point, and collects a large field-of-view image containing the sulfur hexafluoride pressure gauge to be identified. S4: The recognition software module uses a trained YOLO model to recognize the large field-of-view image of the sulfur hexafluoride (SF6) pressure gauge, obtains the position information of the SF6 pressure gauge dial, and adjusts the azimuth and pitch angle of the camera according to the position information to center the SF6 pressure gauge in the image and ensure that the outer contour occupies an area of 1 / 4 to 1 / 2 of the entire image; based on the small field-of-view image of the SF6 pressure gauge to be recognized acquired by the camera; S5: The recognition software module uses the trained YOLO model to recognize the small field-of-view image and obtains the position information and confidence information of the dial, anchor point and pointer of the sulfur hexafluoride pressure gauge; S6: The recognition software module combines the prior information of the instruments at the inspection points to obtain accurate position information of the dial, anchor point and pointer; S7: The identification software module connects the anchor point and pointer center point of the sulfur hexafluoride pressure gauge obtained from S5 to form the actual position information of the pointer, and calculates the angle between the pointer and the zero mark by combining the prior information of the instrument. S8: Calculate the reading of the sulfur hexafluoride pressure gauge based on the prior information of the instruments at the inspection points and the included angle; Wherein: the prior information of the instrument is based on manual calibration.
[0023] In some embodiments, S1, before image processing, further includes: cleaning the pre-collected sulfur hexafluoride pressure gauge images to remove duplicate images, missing images and useless images, and using an image annotation tool to annotate the cleaned images to annotate the dial, anchor point and pointer in the sulfur hexafluoride pressure gauge images. Before sending the labeled sulfur hexafluoride pressure gauge image data into the target detection network YOLO, the process also includes: dividing the shuffled labeled image dataset into training set, validation set and test set according to different ratios (such as 7:2:1).
[0024] In some embodiments, in step S2, the object detection network YOLO is first constructed. The network model specifically includes an input terminal, a backbone, a neck, and a prediction output terminal. The YOLO input terminal performs image preprocessing before inputting the model. Image preprocessing includes image resizing, RGB value normalization of image pixels, and image data enhancement. Image data enhancement increases sample diversity through image processing methods such as random cropping, scaling, rotation, brightness changes, adding Gaussian noise, and randomly stitching multiple images according to a certain ratio. The prediction output terminal is the detection head, composed of convolutional layers, candidate box generation layers, prediction layers, and activation functions, used to generate the final object detection result. Its total loss function is calculated as follows:
[0025] in, For the total loss function, For classifying losses, For confidence loss, To locate the loss, For classification loss weights, For confidence loss weights, To determine the loss weights.
[0026] In some embodiments, in step 2, the hyperparameters of the network model are set, and the model is trained until it converges using the training and validation sets in the labeled dataset. Finally, the YOLO model weights with the best recognition performance are selected by testing on the test set.
[0027] In some embodiments, in step S5, the prior information of the instrument at the inspection point includes the confidence level of the pointer. The confidence level of the pointer is the credibility of the identification result. The confidence level is the credibility of the identification result and has a value of 0-1. The closer it is to 1, the more reliable and accurate the identification result is. The confidence level threshold is set to 0.3. Detection results with a confidence level less than 0.3 are discarded, and detection results with a confidence level greater than or equal to 0.3 are retained.
[0028] In some embodiments, in step S6, the prior information of the instrument at the inspection point includes single dial information. The identification results in step S5 are filtered and grouped, invalid dials are arbitrated and deleted, and the final accurate dial position information is obtained.
[0029] In some embodiments, in step S6, the prior information of the instruments at the inspection points includes anchor points and pointers. The identification results in step S5 are filtered and grouped, and invalid anchor points and pointers are arbitrated and deleted to obtain accurate anchor points and pointers.
[0030] In some embodiments, in step S6, the identification software module combines the prior information of the instrument at the inspection point and groups them according to whether they are within the dial, ultimately obtaining accurate dials, anchor points, and pointers.
[0031] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention. The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the protection scope of the present invention.
Claims
1. A method for reading images from a sulfur hexafluoride pressure gauge, characterized in that, Includes the following steps: S1: Organize the pre-collected sulfur hexafluoride pressure gauge images and use image annotation software to annotate the dial, anchor points, and pointers in the images; S2: Input the labeled sulfur hexafluoride pressure gauge image data into the target detection network YOLO to train the YOLO model; S3: The inspection robot arrives at the inspection point according to the inspection route. The robot adjusts the camera to the pre-set position and pitch angle of the inspection point to take pictures and collect images of the sulfur hexafluoride pressure gauge at the inspection point, and collects a large field-of-view image containing the sulfur hexafluoride pressure gauge to be identified. S4: The recognition software module uses a trained YOLO model to recognize the large field-of-view image of the sulfur hexafluoride (SF6) pressure gauge, obtains the position information of the SF6 pressure gauge dial, and adjusts the azimuth and pitch angle of the camera according to the position information to center the SF6 pressure gauge in the image and ensure that the outer contour occupies an area of 1 / 4 to 1 / 2 of the entire image; based on the small field-of-view image of the SF6 pressure gauge to be recognized acquired by the camera; S5: The recognition software module uses the trained YOLO model to recognize the small field-of-view image and obtains the position information and confidence information of the dial, anchor point and pointer of the sulfur hexafluoride pressure gauge; S6: The recognition software module combines the prior information of the instruments at the inspection points to obtain accurate position information of the dial, anchor point and pointer; S7: The identification software module connects the anchor point and pointer center point of the sulfur hexafluoride pressure gauge obtained from S5 to form the actual position information of the pointer, and calculates the angle between the pointer and the zero mark by combining the prior information of the instrument. S8: Calculate the reading of the sulfur hexafluoride pressure gauge based on the prior information of the instruments at the inspection points and the included angle; Wherein: the prior information of the instrument is based on manual calibration.
2. The method for reading images of a sulfur hexafluoride pressure gauge according to claim 1, characterized in that, In step S1, before image processing, the following steps are also included: cleaning the pre-collected sulfur hexafluoride pressure gauge images to remove duplicate, missing, and useless images, and using an image annotation tool to annotate the cleaned images to annotate the dial, anchor points, and pointers in the sulfur hexafluoride pressure gauge images. Before sending the labeled sulfur hexafluoride pressure gauge image data into the target detection network YOLO, the process also includes: dividing the shuffled labeled image dataset into training set, validation set and test set according to different proportions.
3. The method for reading images of a sulfur hexafluoride pressure gauge according to claim 1, characterized in that, In step S2, the object detection network YOLO is first constructed. The network model specifically includes an input terminal, a backbone, a neck, and a prediction output terminal. The YOLO input terminal performs image preprocessing before inputting the model. Image preprocessing includes image resizing, RGB value normalization of image pixels, and image data augmentation. Image data augmentation increases sample diversity through random cropping, scaling, rotation, brightness changes, adding Gaussian noise, and randomly stitching multiple images together in a certain proportion. The prediction output terminal is the detection head, composed of convolutional layers, candidate box generation layers, prediction layers, and activation functions, used to generate the final object detection result. Its total loss function is calculated as follows: in, For the total loss function, For classifying losses, For confidence loss, To locate the loss, For classification loss weights, For confidence loss weights, To determine the loss weights.
4. The method for reading images of a sulfur hexafluoride pressure gauge according to claim 1, characterized in that, In step 2, the hyperparameters of the network model are set, and the model is trained until it converges using the training and validation sets in the labeled dataset. Finally, the YOLO model weights with the best recognition performance are selected by testing on the test set.
5. The method for reading images of a sulfur hexafluoride pressure gauge according to claim 1, characterized in that, In S5, the prior information of the instrument at the inspection point includes the confidence level of the pointer. The confidence level of the pointer is the credibility of the recognition result. The confidence level threshold is set to 0.
3. Detection results with a confidence level less than 0.3 are discarded, and detection results with a confidence level greater than or equal to 0.3 are retained.
6. The method for reading images of a sulfur hexafluoride pressure gauge according to claim 1, characterized in that: In step S6, the prior information of the instrument at the inspection point includes single dial information. The identification results in step S5 are filtered and grouped, and invalid dials are arbitrated and deleted to obtain the final accurate dial position information.
7. The method for reading images of a sulfur hexafluoride pressure gauge according to claim 1, characterized in that: In step S6, the prior information of the instruments at the inspection points also includes anchor points and pointers. The identification results in step S5 are filtered and grouped, and invalid anchor points and pointers are arbitrated and deleted to obtain accurate anchor points and pointers.
8. The method for reading images of a sulfur hexafluoride pressure gauge according to claim 1, characterized in that: The S6 also includes an identification software module that combines the prior information of the instrument at the inspection point to group the data based on whether it is within the dial, and finally obtains the accurate dial, anchor point and pointer.