Substation equipment component heat defect identification method, system, device and medium

Abstract

The application relates to the technical field of power equipment heating defect identification, and discloses a substation equipment component heating defect identification method, system, device and medium, the method comprising the following steps: inputting an infrared image of a substation equipment into an equipment positioning model to perform image positioning identification, obtaining positioning identification information; performing data processing and grouping on the positioning identification information, obtaining a three-phase grouping result of the substation equipment; obtaining the highest temperature of three phases according to the three-phase grouping result, and performing component heating defect identification on the substation equipment according to the highest temperature of the three phases, obtaining an identification result. The application realizes automatic temperature measurement of the equipment by automatically detecting and grouping the infrared image through infrared shooting of the substation equipment by a unmanned aerial vehicle. The application effectively improves the efficiency and accuracy of substation equipment component heating identification, reduces the labor cost of substation inspection, improves the inspection efficiency, and further guarantees the safety and stability of power facilities.

Description

Technical Field

[0001] This invention relates to the field of power equipment heating defect identification technology, and in particular to a method, system, equipment and medium for identifying heating defects in substation equipment components. Background Technology

[0002] Overheating in substation equipment poses a significant challenge to power system safety and equipment lifespan. During equipment operation or malfunction, energy loss leads to increased internal temperatures. If these temperatures exceed normal ranges, not only will the equipment's lifespan be significantly shortened, but safety hazards may also arise. In extreme cases, overheating can even cause accidents, directly threatening grid stability. Therefore, routine substation inspections must include three-phase temperature monitoring of critical components. Temperature monitoring effectively prevents and promptly addresses overheating issues, ensuring the safe and stable operation of the power system.

[0003] Currently, most substation infrared temperature measurement methods require fixed-point installation of infrared temperature measuring devices and manual assistance, i.e., manual handheld infrared temperature measuring devices or manual judgment of the three phases of equipment. Existing methods have several problems. Firstly, the measurement methods lack flexibility and are difficult to implement. Substation infrared temperature measurement typically uses fixed infrared cameras, ground-based mobile robots, or handheld infrared thermal imagers. Fixed infrared cameras have limited coverage, relatively fixed shooting angles, lack flexibility, and require manual installation; ground-based mobile robots are expensive and difficult to fully capture for high-positioned equipment, such as tubular rigid busbars; handheld infrared thermal imagers require manual inspection and tower climbing. Secondly, there is a lack of methods for automatic three-phase detection and grouping of equipment components. Since bushings and joints of tubular rigid busbars are current-heated devices, the relative temperature difference method is generally used to identify heating defects. Therefore, precise location extraction of bushings and joints and automatic three-phase grouping are required. Current methods mostly involve directly installing infrared temperature measuring devices on the three phases or manual temperature measurement and judgment. Furthermore, existing methods also suffer from the need for manual assistance and low efficiency in inspection and monitoring. Therefore, there is an urgent need for a method that can automatically detect and classify the three phases of equipment in order to achieve accurate, efficient and automated identification of equipment heating defects. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method, system, device, and medium for identifying thermal defects in substation equipment components. This addresses the issues of inflexibility, implementation difficulty, and low efficiency of existing infrared temperature measurement methods for substations, which require manual assistance. The invention aims to reduce labor costs for substation inspections, improve inspection efficiency, and ensure the safety and stability of power facilities.

[0005] In a first aspect, the present invention provides a method for identifying thermal defects in substation equipment components, the method comprising:

[0006] The infrared image of the substation equipment is input into a pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information. The substation equipment is a power equipment with a three-phase joint and bushing structure. The positioning and recognition information includes the overall position information of the three phases, the overall position information of the bushing, the single-phase segmentation information of the bushing, and the single-phase segmentation information of the joint.

[0007] The location identification information is processed and grouped to obtain the three-phase grouping results of the substation equipment. The three-phase grouping results include the three phases of the connector, the three phases of the bushing, and the three phases of the bushing.

[0008] Based on the three-phase grouping results, the highest three-phase temperature is obtained, and the substation equipment is used to identify component heating defects based on the highest three-phase temperature, and the identification results are obtained.

[0009] Furthermore, the step of inputting the infrared image of the substation equipment into a pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information includes:

[0010] The infrared images of the substation equipment are input into the pre-trained three-phase overall positioning sub-model, bushing overall positioning sub-model and single-phase segmentation sub-model for positioning and identification, so as to obtain the three-phase overall position information, bushing overall position information, bushing single-phase segmentation information and joint single-phase segmentation information.

[0011] Furthermore, the step of processing and grouping the positioning and identification information to obtain the three-phase grouping results of the substation equipment includes:

[0012] Based on the crossover ratio between the overall three-phase position information and the overall bushing position information, the overall bushing position information is filtered to obtain an overall bushing set. Then, based on the variance of the ordinate of the geometric center point, the overall bushing position information in the overall bushing set is combined to obtain the overall three-phase bushing information.

[0013] Based on the crossover ratio between the overall location information of the casing and the single-phase segmentation information of the casing, the single-phase segmentation information of the casing is filtered to obtain a single-phase segmentation set of the casing. Then, based on the segmentation area and the relative position of the geometric center point, the single-phase segmentation information of the casing in the single-phase segmentation set of the casing is filtered and divided to obtain the three-phase grouping result of the casing as a whole. The three-phase grouping result includes the three phases on the casing, the three phases on the casing, and the single phase of the casing.

[0014] Furthermore, the step of processing and grouping the positioning and identification information to obtain the three-phase grouping results of the substation equipment includes:

[0015] Based on the crossover ratio between the overall three-phase position information and the single-phase segmentation information of the connector, the single-phase segmentation information of the connector is filtered to obtain a single-phase segmentation set of the connector. Then, based on the variance of the ordinate of the geometric center point, the single-phase segmentation information of the connector in the single-phase segmentation set of the connector is combined to obtain the three phases of the connector in the three-phase whole.

[0016] Further, the step of filtering the overall bushing position information based on the crossover ratio between the overall three-phase position information and the overall bushing position information to obtain an overall bushing set, and combining the overall bushing position information in the overall bushing set according to the variance of the ordinate of the geometric center point to obtain the overall three-phase bushing information includes:

[0017] Calculate the first crossover ratio between each three-phase overall position information and each casing overall position information, compare the first crossover ratio with a first threshold, and filter out the casing overall position information corresponding to the first crossover ratio greater than the first threshold to obtain the casing overall set;

[0018] Determine whether the number of overall position information of the bushing in the overall bushing set is less than or equal to three. If so, the overall position information of the bushing in the overall bushing set is taken as the overall information of the three-phase bushing.

[0019] If not, calculate the ordinate of the geometric center point of the instance box corresponding to each overall position information of the casing in the overall casing set, and divide the overall position information of the casing in the overall casing set into several groups of three;

[0020] Based on the ordinate of the geometric center point, calculate the variance of the ordinate of the geometric center point of each bushing overall position information group, and take the bushing overall position information group with the smallest variance as the overall information of the three-phase bushing.

[0021] Further, the step of filtering the single-phase segmentation information of the casing based on the intersection-union ratio between the overall location information of the casing and the single-phase segmentation information of the casing to obtain a single-phase segmentation set of the casing, and filtering and dividing the single-phase segmentation information of the casing in the single-phase segmentation set according to the segmentation area and the relative position of the geometric center point to obtain the three-phase grouping result of the entire casing includes:

[0022] Calculate the second crossover ratio between the overall position information of each casing and the minimum containment rectangle corresponding to each casing single-phase segmentation information. Compare the second crossover ratio with a second threshold and filter out the casing single-phase segmentation information corresponding to the second crossover ratio greater than the second threshold to obtain the casing single-phase segmentation set.

[0023] The number of single-phase segmentation information in the single-phase segmentation set of the casing is compared with that of two-phase segmentation. If the number is greater than two, the segmentation area of ​​each single-phase segmentation information in the single-phase segmentation set of the casing is calculated. The segmentation areas are sorted in descending order, and the first two single-phase segmentation information in the sorted list are selected to update the single-phase segmentation set of the casing.

[0024] If the quantity is equal to two, then according to the relative position of the geometric center point of the single-phase segmentation information of the single-phase segmentation information of the single-phase segmentation set of the single-phase segmentation, the single-phase segmentation information of the single-phase segmentation ...

[0025] If the number is less than two, then the single-phase segmentation information of the casing in the single-phase segmentation set is taken as the single phase of the casing.

[0026] Further, the step of filtering the single-phase segmentation information of the joint based on the crossover ratio between the overall three-phase position information and the single-phase segmentation information of the joint to obtain a single-phase segmentation set of the joint, and combining the single-phase segmentation information of the joint in the single-phase segmentation set of the joint according to the variance of the ordinate of the geometric center point to obtain the three phases of the joint in the three-phase whole includes:

[0027] Calculate the third crossover ratio between the minimum bounding rectangles corresponding to each three-phase overall position information and each single-phase segmentation information of the connectors, compare the third crossover ratio with a third threshold, and filter out the single-phase segmentation information of the connectors corresponding to the third crossover ratio that is greater than the third threshold to obtain the single-phase segmentation set of the connectors.

[0028] Determine whether the number of single-phase segmentation information in the single-phase segmentation set of the connector is less than or equal to three. If so, then the single-phase segmentation information in the single-phase segmentation set of the connector is regarded as the three phases of the connector in the three-phase whole.

[0029] If not, calculate the ordinate of the geometric center point of the smallest containing rectangle corresponding to the single-phase segmentation information in the single-phase segmentation set of the connector, and divide the single-phase segmentation information in the single-phase segmentation set of the connector into several single-phase segmentation information groups in groups of three.

[0030] Based on the ordinate of the geometric center point, calculate the variance of the ordinate of the geometric center point of each single-phase segmentation information group of the connector, and take the single-phase segmentation information group of the connector with the smallest variance as the three phases of the connector in the three-phase whole.

[0031] In a second aspect, the present invention provides a substation equipment component thermal defect identification system, the system comprising:

[0032] The positioning and recognition module is used to input the infrared image of the substation equipment into a pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information. The substation equipment is a power equipment with a three-phase joint and bushing structure. The positioning and recognition information includes the overall position information of the three phases, the overall position information of the bushing, the single-phase segmentation information of the bushing, and the single-phase segmentation information of the joint.

[0033] The grouping module is used to process and group the positioning and identification information to obtain the three-phase grouping results of the substation equipment. The three-phase grouping results include the three-phase of the connector, the three-phase of the upper bushing, and the three-phase of the lower bushing.

[0034] The heat dissipation identification module is used to obtain the highest temperature of the three phases based on the three-phase grouping results, and to identify the component heat dissipation defects of the substation equipment based on the highest temperature of the three phases, thereby obtaining the identification results.

[0035] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0036] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0037] This invention provides a method, system, equipment, and medium for identifying thermal defects in substation equipment components. By using a drone to take infrared photos of substation equipment and automatically detecting and grouping the infrared images, automatic temperature measurement of the equipment is achieved. This invention can improve the efficiency and accuracy of identifying thermal defects in substation equipment components, reduce the labor costs of substation inspections, improve inspection efficiency, and further ensure the safety and stability of power facilities. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating the method for identifying thermal defects in substation equipment components in an embodiment of the present invention.

[0039] Figure 2 This is a schematic diagram of the structure of the three-phase integrated positioning sub-model in an embodiment of the present invention;

[0040] Figure 3 yes Figure 2 A schematic diagram of the convolutional module in the Backbone;

[0041] Figure 4 yes Figure 2 A schematic diagram of the CSP module in Backbone;

[0042] Figure 5 yes Figure 4 A schematic diagram of the Darknet Bottleneck module;

[0043] Figure 6 yes Figure 2 A schematic diagram of the SPPF module in Backbone;

[0044] Figure 7 yes Figure 2 A schematic diagram of the structure of the Detection Head module;

[0045] Figure 8 This is a schematic diagram of the input and output of the three-phase integrated positioning sub-model in an embodiment of the present invention;

[0046] Figure 9 This is a schematic diagram of the input and output of the overall sleeve positioning sub-model in an embodiment of the present invention;

[0047] Figure 10 This is a schematic diagram of the structure of the Segment Head module of the single-phase segmentation sub-model in this embodiment of the invention;

[0048] Figure 11 This is a schematic diagram of the input and output of the single-phase segmentation sub-model in an embodiment of the present invention;

[0049] Figure 12 This is a schematic diagram of the structure of the substation equipment component heating defect identification system in an embodiment of the present invention;

[0050] Figure 13 This is an internal structural diagram of the computer device in an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Please see Figure 1 The first embodiment of the present invention proposes a method for identifying heating defects in substation equipment components, which includes steps S10 to S30:

[0053] Step S10: Input the infrared image of the substation equipment into the pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information. The substation equipment is a power equipment with a three-phase connector and bushing structure. The positioning and recognition information includes the overall position information of the three phases, the overall position information of the bushing, the single-phase segmentation information of the bushing, and the single-phase segmentation information of the connector.

[0054] Step S20: Perform data processing and grouping on the positioning and identification information to obtain the three-phase grouping results of the substation equipment. The three-phase grouping results include the three phases of the connector, the three phases on the bushing, and the three phases below the bushing.

[0055] Step S30: Based on the three-phase grouping results, the highest temperature of the three phases is obtained, and the component heating defects of the substation equipment are identified based on the highest temperature of the three phases to obtain the identification results.

[0056] This invention provides a method for identifying component overheating defects in substation equipment. Here, substation equipment refers to power equipment with three-phase joints and bushing structures, such as tubular rigid busbars, main transformers, outdoor substation transformers, circuit breakers, disconnect switches, current transformers, and other power equipment.

[0057] First, this invention uses an equipment positioning model to locate and identify infrared images of substation equipment, obtains information on each component of the substation equipment, and automatically groups the information of each component to obtain the various parts of the three-phase system of the substation equipment. The equipment positioning model will be described below.

[0058] In this embodiment, the equipment positioning model includes a three-phase overall positioning sub-model, a bushing overall positioning sub-model, and a single-phase segmentation sub-model. The three-phase overall positioning sub-model is built based on the YOLOv8 model and consists of three parts. Please refer to [link / reference needed]. Figure 2 This includes the Backbone, Neck, and Detection Head.

[0059] Specifically, the backbone consists of a convolutional module, a Cross-Stage Partial (CSP) module, and a Spatial Pyramid Pooling with Feature Fusion (SPPF) module. Here, a network layer consisting of one convolutional module and one CSP module is called a combined network layer. The input infrared image first passes through a convolutional module, then through three of the aforementioned combined network layers, and finally through the SPPF module. The structure of the convolutional module is as follows: Figure 3 As shown, it includes convolutional layers, batch normalization, and activation functions (SiLU).

[0060] The CSP module is used to enhance the representational power of feature maps and improve detection accuracy. Its structure is as follows: Figure 4 As shown, the network includes a convolutional module, a feature splitting module, a Darknet Bottleneck module, and a feature fusion module. The convolutional module in the CSP module has the same structure as the convolutional module in Backbone. The feature splitting module divides the input feature map into two sub-feature maps. The Darknet Bottleneck module reduces computational cost by decreasing the dimensionality of the intermediate feature maps while maintaining the network's representational power; its structure is shown below. Figure 5 As shown, it includes a convolutional module, three max-pooling modules, and a feature fusion module; the feature fusion module of the CSP module is used to combine multiple sub-feature maps obtained by the above modules through skip connections. This helps to retain more spatial information.

[0061] The SPPF module combines the concepts of Spatial Pyramid Pooling (SPP) and Feature Fusion to capture and fuse features at different scales, thereby improving the accuracy of object detection. It pools the input image at different scales and then combines these pooled feature maps to form a fixed-size feature vector for use by subsequent fully connected layers. Its structure is as follows: Figure 6 As shown, it includes a convolution module, three max pooling modules, and a feature combination module.

[0062] The main function of Neck is to fuse feature maps from different layers to capture target information at different scales. This is particularly important for target detection tasks because targets may appear at different scales and locations. Neck mainly includes a feature fusion module, an upsampling module, a convolution module, and a CSP module. The feature fusion module fuses feature maps from different layers, enabling Neck to capture multi-scale information. This is achieved through skip connections, which combine shallow and deep feature maps. The upsampling module is used to recover the spatial dimensions of the feature maps, which helps to preserve the spatial information of the target in the detection head. There are multiple convolution modules used to further extract and refine features. The CSP module has the same function and structure as the CSP module mentioned above, and will not be repeated here.

[0063] The Detection Head is primarily responsible for the two main parts of the object detection task—classification and regression. The structure of the Decoupled Head is as follows: Figure 7 As shown, the network includes two parallel branches: a classification branch and a regression branch, both implemented by convolutional layers. These branches are decoupled and perform predictions independently, rather than simply sharing parameters from the previous layer. This decoupling design helps the network focus more on its respective task, thereby improving detection accuracy.

[0064] For model training, infrared images of substation equipment actually captured by drones are used as the training dataset. Drones equipped with infrared cameras are selected, and efforts are made to capture the entire three phases of the equipment as completely as possible, with the three phases occupying as large a proportion of the image. The images should be clear and stored in a lossless compression format, such as PNG or TIFF.

[0065] The acquired device images are then preprocessed, including Mosaic data augmentation, random translation, random scaling, image size adjustment, Albumentations transformation, random adjustment of image HSV, and random horizontal flipping. Specifically, Mosaic data augmentation involves randomly cropping and arranging four images, then stitching them together to form a new image as training data; random translation randomly moves the image in both width and height; random scaling allows the image to be resized between 50% and 150%, which can be achieved using the `getRotationMatrix2D` function in the OpenCV library; image resizing adjusts the longest side of the image to 640 while maintaining the aspect ratio, which can be achieved using the `resize` and `copyMakeBorder` functions in the OpenCV library; Albumentations transformations include applying blur, median blur, converting to grayscale, contrast-limited adaptive histogram equalization, randomly changing brightness and contrast, random gamma transformation, and reducing image quality through compression, which can be implemented using the `albumentations` image enhancement library in Python; random HSV adjustment applies random HSV enhancements to the image within a predefined range, which can be achieved using the `merge` and `cvtColor` functions in the OpenCV library; and random horizontal flipping refers to horizontally flipping the image with a probability of 0.5, which can be achieved using the `fliplr` function in the NumPy library. These operations help improve model generalization ability, reduce overfitting, improve model performance, save data collection costs, and adapt to different model architectures.

[0066] In this embodiment, the classification loss uses binary cross-entropy loss, and the bounding box regression loss uses CIOU loss and DFL loss. CIOU loss considers not only the intersection-over-union (IoU) ratio between the predicted and ground truth bounding boxes, but also the relative position and aspect ratio differences of the center points. DFL loss is used to calculate the difference between predicted and ground truth feature points. By setting these loss functions, the model can learn more accurate object orientation and angle information. Preferably, the training parameters are set as follows: 32 images per batch; single-card training; pre-loaded parameters trained on MS-COCO object detection; a total of 100 training epochs; and an input image size of 640.

[0067] After the sub-model is trained, the infrared images of the substation equipment captured by the drone are input into the three-phase overall positioning sub-model for positioning and identification. In this embodiment, the tubular rigid busbar of the substation is used as an example for illustration, which can yield the following results: Figure 8 The three-phase overall position information of the tubular rigid busbar is shown.

[0068] The bushing overall positioning sub-model in this embodiment is built based on the YOLOv8 model. Its structure and training method are consistent with the three-phase overall positioning sub-model, and will not be repeated here. The input image of the bushing overall positioning sub-model is also the infrared image of the substation tubular rigid busbar, and the output result is the overall position information of the bushing in the infrared image, i.e., as shown below. Figure 9 As shown.

[0069] The single-phase segmentation sub-model in this embodiment is also built based on the YOLOv8 model. Its sub-model structure includes a Backbone, Neck, and Segment Head. The structures of the Backbone and Neck are consistent with the two sub-models mentioned above and will not be repeated here. Compared to the Detection Head, the Segment Head adds an instance segmentation branch, and its structure is as follows: Figure 10 As shown, the added branches output the segmentation results through a convolutional module.

[0070] For training the single-phase segmentation sub-model, based on the above sub-model training steps, the loss function is supplemented with a binary cross-entropy loss specific to the segmentation. The input and output of the trained single-phase segmentation sub-model are as follows: Figure 11 As shown, the input image is an infrared image of a tubular rigid busbar in a substation, and the output result is the single-phase segmentation information of the bushing and the single-phase segmentation information of the joint of the tubular rigid busbar in the infrared image.

[0071] This invention uses an equipment positioning model to locate and identify infrared images of substation equipment, enabling accurate and efficient identification of the location information of various components of the substation equipment.

[0072] After identifying the overall three-phase location information, overall bushing location information, single-phase bushing segmentation information, and single-phase joint segmentation information of the substation equipment through model positioning, this invention also provides an automatic detection and grouping method to group the positioning and identification information, thereby obtaining the three-phase joints, upper bushing phases, and lower bushing phases belonging to the same three-phase whole in the substation equipment. The specific steps for grouping the upper and lower bushing phases include:

[0073] Based on the crossover ratio between the overall three-phase position information and the overall bushing position information, the overall bushing position information is filtered to obtain an overall bushing set. Then, based on the variance of the ordinate of the geometric center point, the overall bushing position information in the overall bushing set is combined to obtain the overall three-phase bushing information.

[0074] Based on the crossover ratio between the overall location information of the casing and the single-phase segmentation information of the casing, the single-phase segmentation information of the casing is filtered to obtain a single-phase segmentation set of the casing. Then, based on the segmentation area and the relative position of the geometric center point, the single-phase segmentation information of the casing in the single-phase segmentation set of the casing is filtered and divided to obtain the three-phase grouping result of the casing as a whole. The three-phase grouping result includes the three phases on the casing, the three phases on the casing, and the single phase of the casing.

[0075] In this embodiment, firstly, the overall position information of the bushings that may belong to the same three-phase system is filtered from the overall position information of each bushing. Then, three-phase determination is performed on these bushing position information to obtain the overall information of the three-phase bushings that belong to the same three-phase system. The specific steps include:

[0076] Calculate the first crossover ratio between each three-phase overall position information and each casing overall position information, compare the first crossover ratio with a first threshold, and filter out the casing overall position information corresponding to the first crossover ratio greater than the first threshold to obtain the casing overall set;

[0077] Determine whether the number of overall position information of the bushing in the overall bushing set is less than or equal to three. If so, the overall position information of the bushing in the overall bushing set is taken as the overall information of the three-phase bushing.

[0078] If not, calculate the ordinate of the geometric center point of the instance box corresponding to each overall position information of the casing in the overall casing set, and divide the overall position information of the casing in the overall casing set into several groups of three;

[0079] Based on the ordinate of the geometric center point, calculate the variance of the ordinate of the geometric center point of each bushing overall position information group, and take the bushing overall position information group with the smallest variance as the overall information of the three-phase bushing.

[0080] In this embodiment, it is assumed that the output of the three-phase overall positioning sub-model, i.e., the three-phase overall position information, is B. m B mi For B m The i-th predicted output instance; the output of the overall casing positioning sub-model, i.e., the overall casing position information, is B. t B tj For B tThe j-th prediction output instance; the output of the single-phase segmentation sub-model, i.e., the single-phase segmentation information of the bushing, is S. t The single-phase split information of the connector is S. c S tk and S cn S t and S c The kth and nth predicted output instances.

[0081] Then, the above examples are processed and grouped to obtain three bushing systems belonging to the same three-phase system. The specific steps include:

[0082] ① Use cross-combination ratio to screen out the overall set of casings that may belong to the same three-phase main body.

[0083] Based on instance B of each three-phase overall position information mi The overall location information of the bushings is filtered to obtain a set of bushings that may belong to the three-phase system of the equipment. Specifically, for each instance B of the overall location information of the bushings... tj Calculation example B mi With Example B tj The intersection-union ratio, i.e., IoU(B) mi B tj If the intersection-union ratio is greater than the first threshold, then instance B will be... tj Add the casing to the overall assembly φ mti Otherwise, the instance is filtered out. Preferably, the first threshold can be set to 0.9. This step can identify instances that may belong to instance B. mi All instances of the overall location information of the casings, i.e., the overall casing set φ mti Then, the position information of the three bushings belonging to the same three-phase system can be determined from the set, i.e., the following step ②.

[0084] ② Three-phase determination is performed using the variance of the ordinate of the overall geometric center point of the bushing.

[0085] Determine the overall assembly φ of the above-mentioned casing mti If the number of instances contained in the set is less than or equal to three, then the instances in the set are used as the overall information of the three-phase bushing. Otherwise, the three phases are selected by using the variance of the ordinate of the geometric center point of the overall bushing. Specifically, the ordinate of the geometric center point of the instance box corresponding to the position information of each overall bushing in the set is calculated. At the same time, the instances in the set are grouped into groups of three, and the variance of the ordinate of the geometric center point of each group is calculated based on the ordinate of the geometric center point. Then, the group with the smallest variance is used as the overall information of the three-phase bushing belonging to the same three-phase overall.

[0086] In this embodiment, the overall position information of the bushing is filtered and grouped using the cross-comparison ratio and the ordinate of the geometric center point, thereby obtaining the overall information of the three-phase bushing belonging to the same three-phase whole. After obtaining the overall information of the three-phase bushing, it is also necessary to filter and divide the single-phase segmentation information to obtain the upper three phases and lower three phases of the bushing in each bushing whole in the overall information of the three-phase bushing. The specific filtering and division steps are as follows:

[0087] Calculate the second crossover ratio between the overall position information of each casing and the minimum containment rectangle corresponding to each casing single-phase segmentation information. Compare the second crossover ratio with a second threshold and filter out the casing single-phase segmentation information corresponding to the second crossover ratio greater than the second threshold to obtain the casing single-phase segmentation set.

[0088] The number of single-phase segmentation information in the single-phase segmentation set of the casing is compared with that of two-phase segmentation. If the number is greater than two, the segmentation area of ​​each single-phase segmentation information in the single-phase segmentation set of the casing is calculated. The segmentation areas are sorted in descending order, and the first two single-phase segmentation information in the sorted list are selected to update the single-phase segmentation set of the casing.

[0089] If the quantity is equal to two, then according to the relative position of the geometric center point of the single-phase segmentation information of the single-phase segmentation information of the single-phase segmentation set of the single-phase segmentation, the single-phase segmentation information of the single-phase segmentation ...

[0090] If the number is less than two, then the single-phase segmentation information of the casing in the single-phase segmentation set is taken as the single phase of the casing.

[0091] In this embodiment, it is first necessary to filter out all single-phase segmentation information of the bushing that may belong to the same bushing system. Then, these single-phase segmentation information are combined and divided to determine the upper three phases and lower three phases of the bushing system. The specific steps include:

[0092] ③ Use cross-union ratio to screen single-phase segmentation sets of casings that may belong to the same casing unit.

[0093] Example B based on the overall position information of each casing tj The single-phase segmentation information of the bushing is filtered to obtain instance B, which belongs to the overall information of the bushing. tj Single-phase split set φ of the bushing ttj Specifically, for each instance S of single-phase segmentation information of the bushing tk Calculate its minimum containing rectangle B tk Then calculate instance B. tj With instance S tk The crossover ratio between them, i.e., IoU(B) tj Btk If the intersection-union ratio is greater than the second threshold, then instance S will be... tk Add single-phase split set φ to the bushing ttj If not, the instances are filtered out, with the second threshold preferably set to 0.9. This step allows us to obtain single-phase segmentation information for all casing units that may belong to the same casing unit.

[0094] ④ Screening using the single-phase segmentation area of ​​the bushing

[0095] Determine the single-phase split set φ of the bushing ttj If the number of instances in the set is less than or equal to two, and if the number of instances is greater than two, then the segmentation area of ​​the single-phase segmentation information of the bushing is used for comparison and filtering, so that the set φ ttj It contains only two instances; specifically, it calculates the set φ. ttj Instance S of single-phase segmentation information for each bushing tk The segmented areas are arranged in descending order, and only the first two instances in the arrangement are retained.

[0096] ⑤ Determine the three phases above and below the bushing by using the relative positions of the geometric center points of the single-phase bushing segmentation example.

[0097] For a set φ containing only two instances ttj Then, based on the relative positions of the geometric center points of the smallest containing rectangle of the single-phase segmentation information of the bushing, the two instances can be divided, thereby obtaining the upper three phases and lower three phases of the bushing in the whole; if set φ ttj If there is only one instance, then that instance will be output as a single-phase bushing.

[0098] After obtaining the upper three-phase and lower three-phase of the bushing belonging to the same bushing assembly, it is also necessary to obtain the joint three-phase of the corresponding three-phase assembly. The steps for obtaining the joint three-phase of the three-phase assembly through data processing and grouping are as follows:

[0099] Based on the crossover ratio between the overall three-phase position information and the single-phase segmentation information of the connector, the single-phase segmentation information of the connector is filtered to obtain a single-phase segmentation set of the connector. Then, based on the variance of the ordinate of the geometric center point, the single-phase segmentation information of the connector in the single-phase segmentation set of the connector is combined to obtain the three phases of the connector in the three-phase whole.

[0100] In this embodiment, the method for single-phase screening and grouping of connectors is also based on the cross-union ratio and the variance of the ordinate of the geometric center point. The specific steps include:

[0101] Calculate the third crossover ratio between the minimum bounding rectangles corresponding to each three-phase overall position information and each single-phase segmentation information of the connectors, compare the third crossover ratio with a third threshold, and filter out the single-phase segmentation information of the connectors corresponding to the third crossover ratio that is greater than the third threshold to obtain the single-phase segmentation set of the connectors.

[0102] Determine whether the number of single-phase segmentation information in the single-phase segmentation set of the connector is less than or equal to three. If so, then the single-phase segmentation information in the single-phase segmentation set of the connector is regarded as the three phases of the connector in the three-phase whole.

[0103] If not, calculate the ordinate of the geometric center point of the smallest containing rectangle corresponding to the single-phase segmentation information in the single-phase segmentation set of the connector, and divide the single-phase segmentation information in the single-phase segmentation set of the connector into several single-phase segmentation information groups in groups of three.

[0104] Based on the ordinate of the geometric center point, calculate the variance of the ordinate of the geometric center point of each single-phase segmentation information group of the connector, and take the single-phase segmentation information group of the connector with the smallest variance as the three phases of the connector in the three-phase whole.

[0105] In this embodiment, it is first necessary to filter out all single-phase segmentation information of connectors that may belong to the same three-phase system, and then perform three-phase determination on these single-phase segmentation information of connectors to obtain the three phases of the three connectors in the three-phase system. The specific steps are as follows:

[0106] ① Use cross-comparison ratio to screen the single-phase segmentation set of connectors that may belong to the same three-phase system.

[0107] Based on instance B of each three-phase overall position information mi The single-phase segmentation information of the joints is filtered to obtain the set of single-phase segmentations of the joints belonging to the three-phase system. Specifically, for each instance S of single-phase segmentation information of the joints... cn Calculate its minimum containing rectangle B cn And calculate instance B mi With instance S cn The crossover ratio between them, i.e., IoU(B) mi B cn If the intersection-union ratio is greater than the third threshold, then instance S will be... cn Add joint single-phase split set φ mci If not, the instance is filtered out, with the third threshold preferably set to 0.8. This step yields single-phase segmentation information for all connectors that may belong to the same three-phase system.

[0108] ② Using the variance of the ordinate of the geometric center point of the single-phase split joint example to determine the three-phase configuration

[0109] Determine the single-phase split set φ of the connectormci If the number of instances in the set is less than or equal to three, then the instances in the set are output as the three phases of the three-phase system. If not, the three phases are selected using the variance of the ordinate of the geometric center point of the single-phase segmentation information of the joint. Specifically, the set φ is calculated. mci Each instance S in cn The minimum containing rectangle B cn The ordinate of the geometric center point, and the set φ consisting of three points. mci The instances are grouped, and the variance of the geometric center point ordinate of each group is calculated based on the ordinate of the geometric center point. Then, the group with the smallest variance is output as the three-phase connector in the three-phase system.

[0110] After obtaining the three-phase grouping results of the equipment through data processing and grouping, the highest three-phase temperature of the bushing and connector can be identified based on the infrared image corresponding to the three-phase grouping results. Then, the relative temperature difference of the three phases is calculated to achieve temperature output. When the relative temperature difference of the three phases is greater than the threshold, it is considered that there may be a heating defect, and an early warning will be issued. This realizes the identification of heating defects of components of substation equipment. Preferably, in this embodiment, a visual interface can be used to display the relative temperature difference of the three phases and the early warning signal in real time.

[0111] Furthermore, in a preferred embodiment, the present invention also establishes an infrared image database of substation equipment. The data images are infrared images of equipment in a substation scene taken by a drone. Labelme software is used to annotate the three-phase equipment as a whole and the bushings as a whole in the form of minimum containment rectangles. The bushings and connectors are annotated in the form of instance masks, thereby providing data support for subsequent model training and defect identification.

[0112] This embodiment provides a method for identifying overheating defects in substation equipment components. By using a drone to take infrared photos of the substation equipment and automatically detecting and grouping the infrared images, automatic temperature measurement of the equipment is achieved. This invention effectively improves the efficiency and accuracy of overheating identification of substation equipment components, reduces the labor cost of substation inspection, improves inspection efficiency, and further ensures the safety and stability of power facilities.

[0113] Please see Figure 12 Based on the same inventive concept, the second embodiment of the present invention proposes a substation equipment component thermal defect identification system, comprising:

[0114] The positioning and recognition module 10 is used to input the infrared image of the substation equipment into the pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information. The positioning and recognition information includes three-phase overall position information, bushing overall position information, bushing single-phase segmentation information and joint single-phase segmentation information.

[0115] The grouping module 20 is used to process and group the positioning and identification information to obtain the three-phase grouping results of the substation equipment. The three-phase grouping results include the three-phase of the connector, the three-phase of the bushing, and the three-phase of the bushing.

[0116] The heat dissipation identification module 30 is used to obtain the highest temperature of the three phases based on the three-phase grouping results, and to identify the component heat dissipation defects of the substation equipment based on the highest temperature of the three phases, thereby obtaining the identification results.

[0117] The technical features and effects of the substation equipment component thermal defect identification system proposed in this embodiment of the invention are the same as those of the method proposed in this embodiment of the invention, and will not be repeated here. Each module in the above-mentioned substation equipment component thermal defect identification system can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in a computer device in hardware form, or it can be stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0118] Furthermore, embodiments of the present invention also propose a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0119] Please see Figure 13 The diagram illustrates the internal structure of a computer device in one embodiment. This computer device can specifically be a terminal or a server. The computer device includes a processor, memory, network interface, display, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for identifying thermal defects in substation equipment components. The display screen of the computer device can be a liquid crystal display (LCD) or an e-ink display. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0120] Those skilled in the art will understand that Figure 13The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computing devices may include more or fewer components than shown in the diagram, or combine certain components, or have the same component arrangement.

[0121] Furthermore, embodiments of the present invention also propose a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method.

[0122] In summary, this invention proposes a method, system, device, and medium for identifying thermal defects in substation equipment components. The method involves inputting infrared images of the substation equipment into a pre-trained equipment positioning model for image positioning and recognition, obtaining positioning and recognition information. This information includes the overall three-phase position information, the overall bushing position information, single-phase bushing segmentation information, and single-phase joint segmentation information. The positioning and recognition information is then processed and grouped to obtain three-phase grouping results for the substation equipment. These results include three-phase joints, three-phase upper bushings, and three-phase lower bushings. Based on these grouping results, the highest three-phase temperature is determined, and thermal defects in the substation equipment components are identified using these temperatures, yielding the identification results. This invention utilizes a drone to capture infrared images of the substation equipment and automatically detects and groups these images, achieving automatic temperature measurement. This invention effectively improves the efficiency and accuracy of thermal defect identification for substation equipment components, reduces the labor costs of substation inspections, increases inspection efficiency, and further ensures the safety and stability of power facilities.

[0123] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0124] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for identifying thermal defects in substation equipment components, characterized in that, include: The infrared image of the substation equipment is input into a pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information. The substation equipment is a power equipment with a three-phase joint and bushing structure. The positioning and recognition information includes the overall position information of the three phases, the overall position information of the bushing, the single-phase segmentation information of the bushing, and the single-phase segmentation information of the joint. The location identification information is processed and grouped to obtain the three-phase grouping results of the substation equipment. The three-phase grouping results include the three phases of the connector, the three phases above the bushing, and the three phases below the bushing, including: Based on the crossover ratio between the overall three-phase position information and the overall bushing position information, the overall bushing position information is filtered to obtain an overall bushing set. Then, the overall bushing position information in the overall bushing set is combined according to the variance of the ordinate of the geometric center point to obtain the overall three-phase bushing information. Specific steps include: Calculate the first crossover ratio between each three-phase overall position information and each casing overall position information, compare the first crossover ratio with a first threshold, and filter out the casing overall position information corresponding to the first crossover ratio greater than the first threshold to obtain the casing overall set; Determine whether the number of overall position information of the bushing in the overall bushing set is less than or equal to three. If so, the overall position information of the bushing in the overall bushing set is taken as the overall information of the three-phase bushing. If not, calculate the ordinate of the geometric center point of the instance box corresponding to each overall position information of the casing in the overall casing set, and divide the overall position information of the casing in the overall casing set into several groups of three; Based on the ordinate of the geometric center point, calculate the variance of the ordinate of the geometric center point of each bushing overall position information group, and take the bushing overall position information group with the smallest variance as the overall information of the three-phase bushing. Based on the crossover ratio between the overall location information of the bushing and the single-phase segmentation information of the bushing, the single-phase segmentation information of the bushing is filtered to obtain a set of single-phase segmentations of the bushing. Then, based on the segmentation area and the relative position of the geometric center point, the single-phase segmentation information in the set of single-phase segmentations of the bushing is filtered and divided to obtain the three-phase grouping result of the entire bushing. The three-phase grouping result includes the three phases above the bushing, the three phases below the bushing, and the single phase of the bushing. Specific steps include: Calculate the second crossover ratio between the overall position information of each casing and the minimum containment rectangle corresponding to each casing single-phase segmentation information. Compare the second crossover ratio with a second threshold and filter out the casing single-phase segmentation information corresponding to the second crossover ratio greater than the second threshold to obtain the casing single-phase segmentation set. The number of single-phase segmentation information in the single-phase segmentation set of the casing is compared with that of two-phase segmentation. If the number is greater than two, the segmentation area of ​​each single-phase segmentation information in the single-phase segmentation set of the casing is calculated. The segmentation areas are sorted in descending order, and the first two single-phase segmentation information in the sorted list are selected to update the single-phase segmentation set of the casing. If the quantity is equal to two, then according to the relative position of the geometric center point of the single-phase segmentation information of the single-phase segmentation information of the single-phase segmentation set of the single-phase segmentation, the single-phase segmentation information of the single-phase segmentation ... If the number is less than two, then the single-phase segmentation information of the casing in the single-phase segmentation set is taken as the single phase of the casing. Based on the three-phase grouping results, the highest three-phase temperature is obtained, and the substation equipment is used to identify component heating defects based on the highest three-phase temperature, and the identification results are obtained.

2. The method for identifying heating defects in substation equipment components according to claim 1, characterized in that, The step of inputting infrared images of substation equipment into a pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information includes: The infrared images of the substation equipment are input into the pre-trained three-phase overall positioning sub-model, bushing overall positioning sub-model and single-phase segmentation sub-model for positioning and identification, so as to obtain the three-phase overall position information, bushing overall position information, bushing single-phase segmentation information and joint single-phase segmentation information.

3. The method for identifying heating defects in substation equipment components according to claim 1, characterized in that, The step of processing and grouping the positioning and identification information to obtain the three-phase grouping results of the substation equipment includes: Based on the crossover ratio between the overall three-phase position information and the single-phase segmentation information of the connector, the single-phase segmentation information of the connector is filtered to obtain a single-phase segmentation set of the connector. Then, based on the variance of the ordinate of the geometric center point, the single-phase segmentation information of the connector in the single-phase segmentation set of the connector is combined to obtain the three phases of the connector in the three-phase whole.

4. The method for identifying heating defects in substation equipment components according to claim 3, characterized in that, The steps of filtering the single-phase segmentation information of the joint based on the crossover ratio between the overall three-phase position information and the single-phase segmentation information of the joint to obtain a single-phase segmentation set of the joint, and combining the single-phase segmentation information of the joint in the single-phase segmentation set of the joint according to the variance of the ordinate of the geometric center point to obtain the three phases of the joint in the three-phase whole include: Calculate the third crossover ratio between the minimum bounding rectangles corresponding to each three-phase overall position information and each single-phase segmentation information of the connectors, compare the third crossover ratio with a third threshold, and filter out the single-phase segmentation information of the connectors corresponding to the third crossover ratio that is greater than the third threshold to obtain the single-phase segmentation set of the connectors. Determine whether the number of single-phase segmentation information in the single-phase segmentation set of the connector is less than or equal to three. If so, then the single-phase segmentation information in the single-phase segmentation set of the connector is regarded as the three phases of the connector in the three-phase whole. If not, calculate the ordinate of the geometric center point of the smallest containing rectangle corresponding to the single-phase segmentation information in the single-phase segmentation set of the connector, and divide the single-phase segmentation information in the single-phase segmentation set of the connector into several single-phase segmentation information groups in groups of three. Based on the ordinate of the geometric center point, calculate the variance of the ordinate of the geometric center point of each single-phase segmentation information group of the connector, and take the single-phase segmentation information group of the connector with the smallest variance as the three phases of the connector in the three-phase whole.

5. A system for identifying thermal defects in substation equipment components, characterized in that, The system is applied to the method as described in any one of claims 1 to 4, comprising: The positioning and recognition module is used to input the infrared image of the substation equipment into a pre-trained equipment positioning model for image positioning and recognition to obtain positioning and recognition information. The substation equipment is a power equipment with a three-phase joint and bushing structure. The positioning and recognition information includes the overall position information of the three phases, the overall position information of the bushing, the single-phase segmentation information of the bushing, and the single-phase segmentation information of the joint. The grouping module is used to process and group the positioning and identification information to obtain the three-phase grouping results of the substation equipment. The three-phase grouping results include the three-phase of the connector, the three-phase of the upper bushing, and the three-phase of the lower bushing. The heat dissipation identification module is used to obtain the highest temperature of the three phases based on the three-phase grouping results, and to identify the component heat dissipation defects of the substation equipment based on the highest temperature of the three phases, thereby obtaining the identification results.

6. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

Images